{
  "source_commit": "2dc3a78",
  "generated_at": "2026-09-11T00:56:04.735Z",
  "providers": [
    {
      "id": "alibaba-dashscope",
      "name": "Alibaba Model Studio (DashScope)",
      "kind": "first_party",
      "urls": {
        "docs": "https://www.alibabacloud.com/help/en/model-studio/compatibility-of-openai-with-dashscope",
        "console": "https://modelstudio.console.alibabacloud.com"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "DASHSCOPE_API_KEY"
          ],
          "key_prefix": "sk-",
          "extra_headers": {},
          "getting_credentials": "Model Studio console → API Keys. Keys are region-bound: international dashscope-intl vs China dashscope endpoints.",
          "docs": "https://www.alibabacloud.com/help/en/model-studio/first-api-call-to-qwen"
        }
      ],
      "endpoints": [
        {
          "id": "compatible-mode-chat",
          "base_url": "https://dashscope-intl.aliyuncs.com",
          "path": "/compatible-mode/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "anthropic",
          "base_url": "https://dashscope-intl.aliyuncs.com",
          "path": "/apps/anthropic",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "embeddings",
        "batch"
      ],
      "quirks": [
        {
          "text": "China base: https://dashscope.aliyuncs.com/compatible-mode/v1. Soft switches /think and /no_think appended to prompts work on open-source hybrid Qwen3 models and qwen-plus-2025-04-28+ (most recent instruction wins). Open-source hybrid models (qwen3-235b-a22b, qwen3-32b) support streaming only when enable_thinking=true — non-streaming calls error. Thinking-only variants (qwen3-next-80b-a3b-thinking, qwq-plus, kimi-k2-thinking) cannot disable. Regional OpenAI-compatible bases like https://{workspace}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1 also exist; native DashScope endpoint .../api/v1/services/aigc/text-generation/generation with X-DashScope-SSE: enable.",
          "docs": "https://www.alibabacloud.com/help/en/model-studio/deep-thinking"
        },
        {
          "text": "Pay-as-you-go regions: China dashscope.aliyuncs.com, Singapore dashscope-intl, Virginia dashscope-us — each with /compatible-mode/v1 (OpenAI), /apps/anthropic (Anthropic), and native /api/v1. Workspace-dedicated domains https://{WorkspaceId}.{region}.maas.aliyuncs.com are recommended for production; rate-limited trial domains exist (trial.cn-beijing / trial.ap-southeast-1).",
          "docs": "https://help.aliyun.com/en/model-studio/base-url"
        },
        {
          "text": "Temporary API keys: POST /api/v1/tokens (Bearer with permanent key) returns short-lived st- tokens (default 60s, max 1800s) inheriting the parent key's permissions — for browser/mobile use. Alibaba Cloud AccessKey/STS is NOT accepted on inference endpoints.",
          "docs": "https://www.alibabacloud.com/help/en/model-studio/application-obtain-temporary-authentication-token"
        },
        {
          "text": "Savings/resource plans are billing constructs on the same keys and endpoints: AI General-purpose Savings Plan (from $150/mo, discounts up to 32%/47%, no discount for DeepSeek/Kimi/GLM/MiniMax) and LLM Savings Plan prepaid credits (no discount). QwenCloud (qwencloud.com) is a separate-account console using these same endpoints and key formats; Bailian is the China brand of Model Studio.",
          "docs": "https://www.alibabacloud.com/help/en/model-studio/savings-plan-and-resource-package"
        }
      ],
      "offerings": [
        {
          "model": "alibaba/qwen3-7-plus",
          "wire_id": "qwen3.7-plus",
          "endpoint": "anthropic",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.4,
            "output": 1.6,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "notes": "≤256k tier; 256k–1M is 1.20/4.80; cache hits billed at 10% of input (rule-based); limited-time promos may apply.",
            "source": {
              "url": "https://www.alibabacloud.com/help/en/model-studio/model-pricing",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Anthropic-compatible path: thinking defaults ON; enable_thinking controls on this path are not separately documented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://www.alibabacloud.com/help/en/model-studio/deep-thinking",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "alibaba/qwen3-7-plus",
          "wire_id": "qwen3.7-plus",
          "endpoint": "compatible-mode-chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.4,
            "output": 1.6,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "notes": "≤256k tier; 256k–1M is 1.20/4.80; cache hits billed at 10% of input (rule-based); limited-time promos may apply.",
            "source": {
              "url": "https://www.alibabacloud.com/help/en/model-studio/model-pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 65536,
            "source": {
              "url": "https://www.alibabacloud.com/blog/qwen3-7-plus-multimodal-agent-intelligence_603206",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Qwen3.7/3.6/3.5 series default thinking ON. preserve_thinking carries reasoning_content across turns.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://www.alibabacloud.com/help/en/model-studio/deep-thinking",
              "verified": "2026-08-18"
            },
            "toggle": {
              "param": "enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "alibaba/qwen3-8-max",
          "wire_id": "qwen3.8-max",
          "endpoint": "compatible-mode-chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 6,
            "cache_read": 0.25,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://qwen.ai/blog?id=qwen3.8",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://qwen.ai/blog?id=qwen3.8",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Follows the Qwen enable_thinking convention; not documented for qwen3.8-max specifically — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://www.alibabacloud.com/help/en/model-studio/deep-thinking",
              "verified": "2026-08-20"
            },
            "toggle": {
              "param": "enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        }
      ]
    },
    {
      "id": "alibaba-token-plan",
      "name": "Alibaba Token Plan",
      "kind": "subscription",
      "urls": {
        "docs": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview",
        "console": "https://modelstudio.console.alibabacloud.com/",
        "pricing": "https://www.alibabacloud.com/help/en/model-studio/token-plan-personal-overview"
      },
      "auth": [
        {
          "id": "plan-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "BAILIAN_TOKEN_PLAN_API_KEY"
          ],
          "key_prefix": "sk-sp-",
          "extra_headers": {},
          "getting_credentials": "Subscribe in Model Studio (Singapore region); plan keys only work on token-plan endpoints. Env var name not officially documented — verify.",
          "docs": "https://www.alibabacloud.com/help/en/model-studio/token-plan-team-quickstart"
        }
      ],
      "endpoints": [
        {
          "id": "compatible-mode",
          "base_url": "https://token-plan.ap-southeast-1.maas.aliyuncs.com",
          "path": "/compatible-mode/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "anthropic",
          "base_url": "https://token-plan.ap-southeast-1.maas.aliyuncs.com",
          "path": "/apps/anthropic",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "China base: https://token-plan.cn-beijing.maas.aliyuncs.com (same paths). Model allowlist (exact strings): qwen3.8-max-preview (10x credit discount, night pricing 22:00-08:00 UTC+8), qwen3.7-max, qwen3.7-plus, qwen3.6-flash, glm-5.2, deepseek-v4-pro, wan2.7-image(-pro), happyhorse-1.1 video models; Team edition adds Kimi k2.7-code/k2.6/k2.5, GLM 5.1/5, MiniMax-M2.5, qwen3.6-plus/flash.",
          "docs": "https://www.alibabacloud.com/help/en/model-studio/token-plan-personal-overview"
        },
        {
          "text": "Team Edition: seat = one member + one auto-generated key; deduction order seat quota then shared packs then suspension; TPS/TPM limits at primary-account level; RAM policies AliyunTokenPlanFullAccess gate purchase.",
          "docs": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview"
        }
      ],
      "plan": {
        "price_usd": 6,
        "period": "monthly",
        "quota": "Personal: Lite $6 (700 credits/5h, 2500/7d), Standard $20 (3000/10k), Pro $70 (12k/40k). Team: Standard $30/seat (25k credits), Pro $100/seat (100k), Max $200/seat (250k), shared packs $700/625k.",
        "notes": "Sliding 5-hour + 7-day windows, no monthly reset. sk-sp- keys are NOT interchangeable with Coding Plan keys (shared prefix, different product). Interactive-tool use only (Claude Code, Cursor, Qwen Code, OpenClaw); scripts/backends banned on Personal.",
        "docs": "https://www.alibabacloud.com/help/en/model-studio/token-plan-personal-overview"
      },
      "offerings": [
        {
          "model": "deepseek/deepseek-v4-pro",
          "wire_id": "deepseek-v4-pro",
          "endpoint": "compatible-mode",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "DashScope-family enable_thinking toggle; thinking controls not documented for Token Plan endpoints — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://www.alibabacloud.com/help/en/model-studio/deep-thinking",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-7-code",
          "wire_id": "kimi-k2.7-code",
          "endpoint": "compatible-mode",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Team Edition only. DashScope-family enable_thinking toggle; thinking controls not documented for Token Plan endpoints — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://www.alibabacloud.com/help/en/model-studio/deep-thinking",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "zai/glm-5-2",
          "wire_id": "glm-5.2",
          "endpoint": "compatible-mode",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "DashScope-family enable_thinking toggle; thinking controls not documented for Token Plan endpoints — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://www.alibabacloud.com/help/en/model-studio/deep-thinking",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        }
      ]
    },
    {
      "id": "anthropic",
      "name": "Anthropic",
      "kind": "first_party",
      "urls": {
        "docs": "https://platform.claude.com/docs/en/api/overview",
        "console": "https://console.anthropic.com",
        "status": "https://status.anthropic.com",
        "pricing": "https://platform.claude.com/docs/en/docs/about-claude/models"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "x-api-key",
          "env": [
            "ANTHROPIC_API_KEY"
          ],
          "key_prefix": "sk-ant-api",
          "extra_headers": {
            "anthropic-version": "2023-06-01"
          },
          "getting_credentials": "Console → Settings → API Keys. Keys are org-scoped; optional expiry (3h to never).",
          "docs": "https://platform.claude.com/docs/en/manage-claude/authentication"
        },
        {
          "id": "oauth",
          "type": "oauth",
          "transport": "header",
          "header": "x-api-key",
          "env": [],
          "extra_headers": {
            "anthropic-beta": "oauth-2025-04-20",
            "anthropic-version": "2023-06-01"
          },
          "getting_credentials": "Run Claude Code and sign in with a Claude subscription; OAuth tokens are sent via x-api-key (not Bearer).",
          "docs": "https://code.claude.com/docs/en/authentication",
          "flow": "authorization_code_pkce",
          "scopes": [
            "org:create_api_key",
            "user:profile",
            "user:inference"
          ],
          "token_transport": "header"
        }
      ],
      "endpoints": [
        {
          "id": "v1-messages",
          "base_url": "https://api.anthropic.com",
          "path": "/v1/messages",
          "protocol": "anthropic-messages"
        },
        {
          "id": "v1-messages-batches",
          "base_url": "https://api.anthropic.com",
          "path": "/v1/messages/batches",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "batch",
        "count_tokens",
        "prompt_caching",
        "files"
      ],
      "quirks": [
        {
          "text": "Thinking is incompatible with temperature/top_p/top_k, forced tool use, and prefill; changing thinking params invalidates prompt caches; thinking signatures must be replayed unmodified in tool-use turns.",
          "docs": "https://platform.claude.com/docs/en/docs/build-with-claude/extended-thinking"
        }
      ],
      "offerings": [
        {
          "model": "anthropic/claude-fable-5",
          "wire_id": "claude-fable-5",
          "endpoint": "v1-messages",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 10,
            "output": 50,
            "cache_read": 1,
            "cache_write": 12.5,
            "free": false,
            "source": {
              "url": "https://platform.claude.com/docs/en/docs/about-claude/models",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 128000,
            "source": {
              "url": "https://www.anthropic.com/news/claude-fable-5",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": true,
            "default": "adaptive",
            "notes": "thinking.type disabled→400 on fable-5 (mandatory=true there); opus-5/sonnet-5 accept disabled only at effort ≤ high; enabled+budget_tokens→400 on all 5-series; effort changes invalidate prompt-cache breakpoints; 300k output via Batch beta header output-300k-2026-03-24 (opus/sonnet).",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k",
              "prefill"
            ],
            "source": {
              "url": "https://platform.claude.com/docs/en/docs/build-with-claude/adaptive-thinking",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "output_config.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-haiku-4-5",
          "wire_id": "claude-haiku-4-5",
          "endpoint": "v1-messages",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": null,
            "output": null,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://platform.claude.com/docs/en/docs/about-claude/models",
              "verified": "2026-08-18"
            }
          },
          "limits": {
            "context": 200000,
            "output": null,
            "source": {
              "url": "https://platform.claude.com/docs/en/docs/about-claude/models",
              "verified": "2026-08-18"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "budget",
            "mandatory": false,
            "default": "on",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k"
            ],
            "source": {
              "url": "https://platform.claude.com/docs/en/docs/build-with-claude/extended-thinking",
              "verified": "2026-08-18"
            },
            "budget": {
              "param": "thinking.budget_tokens",
              "min": 1024,
              "zero_means_off": false,
              "special_values": {},
              "constraint": "must be < max_tokens"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-5",
          "wire_id": "claude-opus-5",
          "endpoint": "v1-messages",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": 6.25,
            "free": false,
            "source": {
              "url": "https://platform.claude.com/docs/en/docs/about-claude/models",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 128000,
            "source": {
              "url": "https://www.anthropic.com/news/claude-opus-5",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "adaptive",
            "notes": "thinking.type disabled→400 on fable-5 (mandatory=true there); opus-5/sonnet-5 accept disabled only at effort ≤ high; enabled+budget_tokens→400 on all 5-series; effort changes invalidate prompt-cache breakpoints; 300k output via Batch beta header output-300k-2026-03-24 (opus/sonnet).",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k",
              "prefill"
            ],
            "source": {
              "url": "https://platform.claude.com/docs/en/docs/build-with-claude/adaptive-thinking",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "output_config.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-sonnet-5",
          "wire_id": "claude-sonnet-5",
          "endpoint": "v1-messages",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 10,
            "cache_read": 0.2,
            "cache_write": 2.5,
            "free": false,
            "source": {
              "url": "https://platform.claude.com/docs/en/docs/about-claude/models",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 128000,
            "source": {
              "url": "https://www.anthropic.com/news/claude-sonnet-5",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "adaptive",
            "notes": "thinking.type disabled→400 on fable-5 (mandatory=true there); opus-5/sonnet-5 accept disabled only at effort ≤ high; enabled+budget_tokens→400 on all 5-series; effort changes invalidate prompt-cache breakpoints; 300k output via Batch beta header output-300k-2026-03-24 (opus/sonnet).",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k",
              "prefill"
            ],
            "source": {
              "url": "https://platform.claude.com/docs/en/docs/build-with-claude/adaptive-thinking",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "output_config.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        }
      ]
    },
    {
      "id": "aws-bedrock",
      "name": "AWS Bedrock",
      "kind": "cloud_hosted",
      "urls": {
        "docs": "https://docs.aws.amazon.com/bedrock/latest/userguide/api-keys.html",
        "console": "https://console.aws.amazon.com/bedrock"
      },
      "auth": [
        {
          "id": "bearer-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "AWS_BEARER_TOKEN_BEDROCK"
          ],
          "extra_headers": {},
          "getting_credentials": "Bedrock console → API keys. Short-term (<=12h) or long-lived; governed by bedrock:CallWithBearerToken. Not valid for bidirectional streaming, Agents, or Data Automation.",
          "docs": "https://docs.aws.amazon.com/bedrock/latest/userguide/api-keys.html"
        },
        {
          "id": "sigv4",
          "type": "sigv4",
          "transport": "request_signing",
          "env": [
            "AWS_ACCESS_KEY_ID",
            "AWS_SECRET_ACCESS_KEY",
            "AWS_SESSION_TOKEN"
          ],
          "extra_headers": {},
          "getting_credentials": "Standard AWS credentials chain (IAM user, role, SSO). Signs the request with SigV4; service 'bedrock'.",
          "docs": "https://docs.aws.amazon.com/bedrock/latest/userguide/security_iam.html"
        }
      ],
      "endpoints": [
        {
          "id": "converse",
          "base_url": "https://bedrock-runtime.{region}.amazonaws.com",
          "path": "/model/{modelId}/converse",
          "protocol": "bedrock-converse",
          "auth": "sigv4"
        },
        {
          "id": "invoke-anthropic",
          "base_url": "https://bedrock-runtime.{region}.amazonaws.com",
          "path": "/model/{modelId}/invoke",
          "protocol": "anthropic-messages",
          "auth": "sigv4"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "batch",
        "embeddings"
      ],
      "quirks": [
        {
          "text": "Cross-region inference profiles prefix model IDs with us./eu./apac./global. — e.g. us.anthropic.claude-sonnet-5. Streaming required when max_tokens > 21,333 with thinking.",
          "docs": "https://docs.aws.amazon.com/bedrock/latest/userguide/inference-profiles-support.html"
        }
      ],
      "offerings": [
        {
          "model": "anthropic/claude-fable-5",
          "wire_id": "anthropic.claude-fable-5",
          "endpoint": "invoke-anthropic",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 10,
            "output": 50,
            "cache_read": 1,
            "cache_write": 12.5,
            "free": false,
            "notes": "US East on-demand.",
            "source": {
              "url": "https://aws.amazon.com/bedrock/pricing/",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 128000,
            "source": {
              "url": "https://aws.amazon.com/bedrock/claude/",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": true,
            "default": "adaptive",
            "notes": "Adaptive thinking clone of the native fable-5 offering; thinking cannot be disabled (mandatory=true). enabled+budget_tokens→400 on all 5-series; effort changes invalidate prompt-cache breakpoints. Bedrock 5-series IDs carry no -v1:0 suffix; mantle Messages endpoint exists at bedrock-mantle.{region}.api.aws/anthropic/v1/messages.",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k",
              "prefill"
            ],
            "source": {
              "url": "https://docs.aws.amazon.com/bedrock/latest/userguide/claude-messages-extended-thinking.html",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "output_config.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-haiku-4-5",
          "wire_id": "anthropic.claude-haiku-4-5-20251001-v1:0",
          "endpoint": "invoke-anthropic",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1,
            "output": 5,
            "cache_read": 0.1,
            "cache_write": 1.25,
            "free": false,
            "notes": "US East on-demand.",
            "source": {
              "url": "https://aws.amazon.com/bedrock/pricing/",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "budget",
            "mandatory": false,
            "default": "on",
            "notes": "Extended thinking via the native Anthropic surface; budget clone of the Bedrock sonnet-4-6 offering.",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k"
            ],
            "source": {
              "url": "https://docs.aws.amazon.com/bedrock/latest/userguide/claude-messages-extended-thinking.html",
              "verified": "2026-08-19"
            },
            "budget": {
              "param": "thinking.budget_tokens",
              "min": 1024,
              "zero_means_off": false,
              "special_values": {},
              "constraint": "must be < max_tokens"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-5",
          "wire_id": "anthropic.claude-opus-5",
          "endpoint": "invoke-anthropic",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": 6.25,
            "free": false,
            "notes": "US East on-demand.",
            "source": {
              "url": "https://aws.amazon.com/bedrock/pricing/",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 128000,
            "source": {
              "url": "https://aws.amazon.com/bedrock/claude/",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "adaptive",
            "notes": "Adaptive thinking clone of the native opus-5 offering. thinking.type disabled→400 on fable-5; opus-5/sonnet-5 accept disabled only at effort ≤ high; enabled+budget_tokens→400 on all 5-series; effort changes invalidate prompt-cache breakpoints. Bedrock 5-series IDs carry no -v1:0 suffix; mantle Messages endpoint exists at bedrock-mantle.{region}.api.aws/anthropic/v1/messages.",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k",
              "prefill"
            ],
            "source": {
              "url": "https://docs.aws.amazon.com/bedrock/latest/userguide/claude-messages-extended-thinking.html",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "output_config.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-sonnet-5",
          "wire_id": "anthropic.claude-sonnet-5",
          "endpoint": "invoke-anthropic",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 10,
            "cache_read": 0.2,
            "cache_write": 2.5,
            "free": false,
            "notes": "US East on-demand.",
            "source": {
              "url": "https://aws.amazon.com/bedrock/pricing/",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 128000,
            "source": {
              "url": "https://aws.amazon.com/bedrock/claude/",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "adaptive",
            "notes": "Adaptive thinking clone of the native sonnet-5 offering. thinking.type disabled→400 on fable-5; opus-5/sonnet-5 accept disabled only at effort ≤ high; enabled+budget_tokens→400 on all 5-series; effort changes invalidate prompt-cache breakpoints. Bedrock 5-series IDs carry no -v1:0 suffix; mantle Messages endpoint exists at bedrock-mantle.{region}.api.aws/anthropic/v1/messages.",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k",
              "prefill"
            ],
            "source": {
              "url": "https://docs.aws.amazon.com/bedrock/latest/userguide/claude-messages-extended-thinking.html",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "output_config.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        }
      ]
    },
    {
      "id": "azure-foundry",
      "name": "Azure AI Foundry",
      "kind": "cloud_hosted",
      "urls": {
        "docs": "https://learn.microsoft.com/en-us/azure/foundry/openai/api-version-lifecycle",
        "console": "https://ai.azure.com"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "api-key",
          "env": [
            "AZURE_OPENAI_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Foundry resource → Keys and Endpoint. 32-char hex keys.",
          "docs": "https://learn.microsoft.com/en-us/azure/foundry/openai/api-version-lifecycle"
        },
        {
          "id": "entra",
          "type": "entra_bearer",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [],
          "extra_headers": {},
          "getting_credentials": "Microsoft Entra token with scope https://ai.azure.com/.default; SDKs refresh automatically. Roles like 'Cognitive Services OpenAI User'.",
          "docs": "https://learn.microsoft.com/en-us/azure/foundry/openai/api-version-lifecycle"
        }
      ],
      "endpoints": [
        {
          "id": "v1-responses",
          "base_url": "https://{resource}.openai.azure.com",
          "path": "/openai/v1/responses",
          "protocol": "openai-responses"
        },
        {
          "id": "v1-chat-completions",
          "base_url": "https://{resource}.openai.azure.com",
          "path": "/openai/v1/chat/completions",
          "protocol": "openai-chat"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "embeddings",
        "files",
        "batch",
        "fine_tuning"
      ],
      "quirks": [
        {
          "text": "The v1 route needs no api-version query param and works with unmodified OpenAI SDKs (set OPENAI_BASE_URL). Classic route uses deployment names + api-version; responses include content_filter_results.",
          "docs": "https://learn.microsoft.com/en-us/azure/foundry/openai/api-version-lifecycle"
        }
      ],
      "offerings": [
        {
          "model": "openai/gpt-5-6",
          "wire_id": "gpt-5.6",
          "endpoint": "v1-responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 12,
            "cache_read": 0.2,
            "cache_write": 2.5,
            "free": false,
            "notes": "Bills as the Terra tier; variant prices: sol 5/30, luna 0.2/1.2, tiered over 200k — verify mapping.",
            "source": {
              "url": "https://azure.microsoft.com/en-us/blog/gpt-5-6-now-available-in-microsoft-foundry/",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Same per-model matrix as OpenAI: none default on 5.1, xhigh on 5.4+, max on 5.6 Responses.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p",
              "max_tokens"
            ],
            "source": {
              "url": "https://learn.microsoft.com/en-us/azure/foundry/openai/how-to/reasoning",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        }
      ]
    },
    {
      "id": "baseten",
      "name": "Baseten",
      "kind": "first_party",
      "urls": {
        "docs": "https://docs.baseten.co/inference/model-apis/overview",
        "console": "https://www.baseten.co",
        "pricing": "https://www.baseten.co/pricing/"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "BASETEN_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create keys in the Baseten dashboard (Settings > API Keys). Send as Authorization: Bearer BASETEN_API_KEY; the legacy Api-Key header scheme is also accepted.",
          "docs": "https://docs.baseten.co/inference/model-apis/overview"
        }
      ],
      "endpoints": [
        {
          "id": "chat-completions",
          "base_url": "https://inference.baseten.co",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "messages",
          "base_url": "https://inference.baseten.co",
          "path": "/v1/messages",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Baseten reads Authorization only — the Anthropic SDK's default x-api-key is NOT read; override default_headers.",
          "docs": "https://docs.baseten.co/inference/model-apis/overview"
        },
        {
          "text": "reasoning_effort is validated (invalid values 400); chat_template_args.enable_thinking on opt-in models is unvalidated and fails silently.",
          "docs": "https://docs.baseten.co/inference/model-apis/reasoning"
        },
        {
          "text": "Rate limits are account-wide (Basic verified 120 RPM / 500k TPM); 529 = overloaded; GET /v1/models embeds pricing and context.",
          "docs": "https://docs.baseten.co/inference/model-apis/pricing-and-limits"
        },
        {
          "text": "The Anthropic-compatible /v1/messages endpoint is in beta; behavior may change before general availability — prefer Chat Completions for production workloads.",
          "docs": "https://docs.baseten.co/inference/model-apis/overview"
        }
      ],
      "offerings": [
        {
          "model": "deepseek/deepseek-v4-flash",
          "wire_id": "deepseek-ai/DeepSeek-V4-Flash-0731",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.13,
            "output": 0.26,
            "cache_read": 0.028,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://www.baseten.co/pricing/",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://docs.baseten.co/inference/model-apis/overview",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.baseten.co/inference/model-apis/reasoning",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "minimal",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "moonshot/kimi-k3",
          "wire_id": "moonshotai/Kimi-K3",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 3,
            "output": 15,
            "cache_read": 0.3,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://www.baseten.co/pricing/",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://docs.baseten.co/inference/model-apis/overview",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.baseten.co/inference/model-apis/reasoning",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "high",
                "max"
              ],
              "default": "max",
              "notes": "none disables reasoning."
            }
          }
        },
        {
          "model": "openai/gpt-oss-120b",
          "wire_id": "openai/gpt-oss-120b",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.1,
            "output": 0.5,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://www.baseten.co/pricing/",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 131072,
            "output": null,
            "source": {
              "url": "https://docs.baseten.co/inference/model-apis/overview",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.baseten.co/inference/model-apis/reasoning",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "thinkingmachines/inkling",
          "wire_id": "thinkingmachines/inkling",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1,
            "output": 4.05,
            "cache_read": 0.17,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://www.baseten.co/pricing/",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 32768,
            "source": {
              "url": "https://docs.baseten.co/inference/model-apis/overview",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Billed as completion tokens; enable_thinking unvalidated per Baseten docs.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.baseten.co/inference/model-apis/reasoning",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "high"
              ],
              "default": "high",
              "notes": "Default high; none disables reasoning."
            }
          }
        },
        {
          "model": "thinkingmachines/inkling-small",
          "wire_id": "thinkingmachines/inkling-small",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.5,
            "output": 1.2,
            "cache_read": 0.1,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://www.baseten.co/pricing/",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 32768,
            "source": {
              "url": "https://docs.baseten.co/inference/model-apis/overview",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Billed as completion tokens; enable_thinking unvalidated per Baseten docs.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.baseten.co/inference/model-apis/reasoning",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "high"
              ],
              "default": "high",
              "notes": "Default high; none disables reasoning."
            }
          }
        },
        {
          "model": "zai/glm-5-2",
          "wire_id": "zai-org/GLM-5.2",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.4,
            "output": 4.4,
            "cache_read": 0.14,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://www.baseten.co/pricing/",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 128000,
            "source": {
              "url": "https://docs.baseten.co/inference/model-apis/overview",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.baseten.co/inference/model-apis/reasoning",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        }
      ]
    },
    {
      "id": "cohere",
      "name": "Cohere",
      "kind": "first_party",
      "urls": {
        "docs": "https://docs.cohere.com/docs/models",
        "console": "https://dashboard.cohere.com",
        "pricing": "https://www.cohere.com/pricing"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "COHERE_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create a key at dashboard.cohere.com; the same key works on the native v2 API and the OpenAI-compatibility surface.",
          "docs": "https://docs.cohere.com/docs/models"
        }
      ],
      "endpoints": [
        {
          "id": "compat-chat-completions",
          "base_url": "https://api.cohere.ai",
          "path": "/compatibility/v1/chat/completions",
          "protocol": "openai-chat"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "embeddings",
        "rerank"
      ],
      "quirks": [
        {
          "text": "Native v2 API at api.cohere.com (different request shape); the OpenAI-compat surface at api.cohere.ai/compatibility/v1 supports chat, embeddings, audio transcriptions — reasoning_effort only none|high there; no store/logit_bias/n.",
          "docs": "https://docs.cohere.com/v2"
        },
        {
          "text": "North is Cohere's enterprise agentic platform (not a model API). Embed/Rerank billed via Model Vault instances.",
          "docs": "https://www.cohere.com/north"
        },
        {
          "text": "Aya vision/translate variants and 8B Ayas retired 2026-04-04.",
          "docs": "https://docs.cohere.com/release-notes"
        }
      ],
      "offerings": [
        {
          "model": "cohere/command-a-plus",
          "wire_id": "command-a-plus-05-2026",
          "endpoint": "compat-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 131072,
            "output": 65536,
            "source": {
              "url": "https://docs.cohere.com/docs/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "none",
            "mandatory": false,
            "default": "off",
            "notes": "Non-reasoning model; free until rate limits on Cohere's platform, so no per-token cost cataloged.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.cohere.com/docs/models",
              "verified": "2026-08-20"
            }
          }
        },
        {
          "model": "cohere/command-a-reasoning",
          "wire_id": "command-a-reasoning-08-2025",
          "endpoint": "compat-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 262144,
            "output": 32768,
            "source": {
              "url": "https://docs.cohere.com/docs/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Compat maps reasoning_effort none|high to the native thinking parameter (budget_tokens on the native v2 API); compat return shape for thinking not documented — verify. Free until rate limits; no published PAYG price.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.cohere.com/docs/command-a-reasoning",
              "verified": "2026-08-20"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "high"
              ],
              "default": "high"
            }
          }
        }
      ]
    },
    {
      "id": "deepseek",
      "name": "DeepSeek",
      "kind": "first_party",
      "urls": {
        "docs": "https://api-docs.deepseek.com/",
        "console": "https://platform.deepseek.com"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "DEEPSEEK_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create a key at platform.deepseek.com. Works in Claude Code, Copilot, OpenCode via the Anthropic-compatible endpoint.",
          "docs": "https://api-docs.deepseek.com/"
        }
      ],
      "endpoints": [
        {
          "id": "chat-completions",
          "base_url": "https://api.deepseek.com",
          "path": "/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "anthropic",
          "base_url": "https://api.deepseek.com",
          "path": "/anthropic",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Root path /chat/completions works without /v1. Stateless OpenAI Responses API also supported. In tool-call flows, prior-turn reasoning_content must be passed back or you get a 400; without tools it is ignored. When thinking is on, temperature/top_p/presence_penalty/frequency_penalty are silently ignored. FIM only in non-thinking mode. The old deepseek-chat / deepseek-reasoner split is gone — same model names serve both modes via the thinking toggle.",
          "docs": "https://api-docs.deepseek.com/guides/thinking_mode"
        }
      ],
      "offerings": [
        {
          "model": "deepseek/deepseek-v4-flash",
          "wire_id": "deepseek-v4-flash",
          "endpoint": "anthropic",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.44,
            "output": 1.32,
            "cache_read": 0.014,
            "cache_write": null,
            "free": false,
            "notes": "Same rates as the native endpoint (no separate /anthropic pricing published). Peak hours; off-peak is half. Cache hit = input cache-read.",
            "source": {
              "url": "https://api-docs.deepseek.com/quick_start/pricing",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Anthropic-compatible surface (for Claude Code et al.): thinking {type} toggle; reasoning_content round-trip applies in tool flows.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://api-docs.deepseek.com/",
              "verified": "2026-08-18"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-flash",
          "wire_id": "deepseek-v4-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.44,
            "output": 1.32,
            "cache_read": 0.014,
            "cache_write": null,
            "free": false,
            "notes": "Peak hours; off-peak is half. Cache hit = input cache-read.",
            "source": {
              "url": "https://api-docs.deepseek.com/quick_start/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 384000,
            "source": {
              "url": "https://api-docs.deepseek.com/quick_start/pricing",
              "verified": "2026-08-18"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Thinking enabled by default, default effort high. OpenAI-style reasoning_effort values map low->low, medium->high, high->high, xhigh->high, max->max; native values low/high/max.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://api-docs.deepseek.com/guides/thinking_mode",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "high",
                "max"
              ],
              "default": "high"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-pro",
          "wire_id": "deepseek-v4-pro",
          "endpoint": "anthropic",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.32,
            "output": 3.96,
            "cache_read": 0.044,
            "cache_write": null,
            "free": false,
            "notes": "Same rates as the native endpoint (no separate /anthropic pricing published). Peak hours; off-peak is half. Cache hit = input cache-read.",
            "source": {
              "url": "https://api-docs.deepseek.com/quick_start/pricing",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Anthropic-compatible surface (for Claude Code et al.): thinking {type} toggle; reasoning_content round-trip applies in tool flows.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://api-docs.deepseek.com/",
              "verified": "2026-08-18"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-pro",
          "wire_id": "deepseek-v4-pro",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.32,
            "output": 3.96,
            "cache_read": 0.044,
            "cache_write": null,
            "free": false,
            "notes": "Peak hours; off-peak is half. Cache hit = input cache-read.",
            "source": {
              "url": "https://api-docs.deepseek.com/quick_start/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 384000,
            "source": {
              "url": "https://api-docs.deepseek.com/quick_start/pricing",
              "verified": "2026-08-18"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Thinking enabled by default, default effort high. OpenAI-style reasoning_effort values map low->low, medium->high, high->high, xhigh->high, max->max; native values low/high/max.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://api-docs.deepseek.com/guides/thinking_mode",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "high",
                "max"
              ],
              "default": "high"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        }
      ]
    },
    {
      "id": "fireworks-ai",
      "name": "Fireworks AI",
      "kind": "first_party",
      "urls": {
        "docs": "https://docs.fireworks.ai/",
        "console": "https://app.fireworks.ai",
        "pricing": "https://docs.fireworks.ai/serverless/pricing"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "FIREWORKS_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create keys in the Fireworks console (app.fireworks.ai, Account Settings > API Keys).",
          "docs": "https://docs.fireworks.ai/"
        }
      ],
      "endpoints": [
        {
          "id": "inference-v1",
          "base_url": "https://api.fireworks.ai",
          "path": "/inference/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "responses",
          "base_url": "https://api.fireworks.ai",
          "path": "/inference/v1/responses",
          "protocol": "openai-responses"
        },
        {
          "id": "anthropic",
          "base_url": "https://api.fireworks.ai",
          "path": "/inference/v1/messages",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "embeddings",
        "batch",
        "rerank"
      ],
      "quirks": [
        {
          "text": "reasoning_effort (low/medium/high) AND an Anthropic-style thinking object {type: enabled, budget_tokens >= 1024} are both accepted on chat/completions — specifying both errors.",
          "docs": "https://docs.fireworks.ai/guides/reasoning"
        },
        {
          "text": "Interleaved thinking: you MUST echo prior assistant reasoning_content when the last message is a tool result; reasoning_history: preserved retains reasoning across turns.",
          "docs": "https://docs.fireworks.ai/guides/reasoning"
        },
        {
          "text": "Batch is 50% of serverless price via the Fireworks-native batch API (not OpenAI /v1/batches); Responses API stores by default; streaming includes usage in the final chunk.",
          "docs": "https://docs.fireworks.ai/guides/reasoning"
        }
      ],
      "offerings": [
        {
          "model": "deepseek/deepseek-v4-pro",
          "wire_id": "deepseek-v4-pro",
          "endpoint": "inference-v1",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.74,
            "output": 3.48,
            "cache_read": 0.145,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://docs.fireworks.ai/serverless/pricing",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://api.fireworks.ai/inference/v1/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.fireworks.ai/guides/reasoning",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-6",
          "wire_id": "kimi-k2p6",
          "endpoint": "inference-v1",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.95,
            "output": 4,
            "cache_read": 0.16,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://docs.fireworks.ai/serverless/pricing",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://api.fireworks.ai/inference/v1/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "reasoning_effort and thinking.budget_tokens are mutually exclusive — specifying both raises a validation error.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.fireworks.ai/guides/reasoning",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            },
            "budget": {
              "param": "thinking.budget_tokens",
              "min": 1024,
              "zero_means_off": false,
              "special_values": {}
            }
          }
        },
        {
          "model": "openai/gpt-oss-120b",
          "wire_id": "gpt-oss-120b",
          "endpoint": "inference-v1",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.15,
            "output": 0.6,
            "cache_read": 0.015,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://docs.fireworks.ai/serverless/pricing",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.fireworks.ai/guides/reasoning",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "none is not offered on this surface."
            }
          }
        },
        {
          "model": "zai/glm-4-7",
          "wire_id": "glm-4p7",
          "endpoint": "inference-v1",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.6,
            "output": 2.2,
            "cache_read": 0.3,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://docs.fireworks.ai/serverless/pricing",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 200000,
            "output": 128000,
            "source": {
              "url": "https://docs.z.ai/guides/llm/glm-4.7",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Per-model reasoning controls not documented; the reasoning guide demos GLM-4.7 with reasoning_history: preserved — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.fireworks.ai/guides/reasoning",
              "verified": "2026-08-19"
            }
          }
        }
      ]
    },
    {
      "id": "google-gemini",
      "name": "Google Gemini API",
      "kind": "first_party",
      "urls": {
        "docs": "https://ai.google.dev/api",
        "console": "https://aistudio.google.com"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "x-goog-api-key",
          "env": [
            "GEMINI_API_KEY",
            "GOOGLE_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "AI Studio → Get API key. Prefer the x-goog-api-key header; the legacy ?key= query param leaks keys into logs.",
          "docs": "https://ai.google.dev/api"
        }
      ],
      "endpoints": [
        {
          "id": "generate-content",
          "base_url": "https://generativelanguage.googleapis.com",
          "path": "/v1beta/models/{model}:generateContent",
          "protocol": "google-generate-content"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "embeddings",
        "files",
        "batch",
        "count_tokens",
        "prompt_caching"
      ],
      "quirks": [
        {
          "text": "OpenAI-compatible route: use https://generativelanguage.googleapis.com/v1beta/openai/ as base URL with the same API key.",
          "docs": "https://ai.google.dev/gemini-api/docs/openai"
        }
      ],
      "offerings": [
        {
          "model": "google/gemini-3-1-pro",
          "wire_id": "gemini-3.1-pro-preview",
          "endpoint": "generate-content",
          "status": "preview",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 12,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "notes": "≤200K tier; >200K doubles",
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/pricing",
              "verified": "2026-08-21"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 65536,
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/models",
              "verified": "2026-08-21"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "thinkingLevel replaces thinkingBudget on Gemini 3; cannot disable thinking.",
            "returns": "thought_parts",
            "must_round_trip": "thought_signature",
            "incompatible_with": [],
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/thinking",
              "verified": "2026-08-21"
            },
            "effort": {
              "param": "generationConfig.thinkingConfig.thinkingLevel",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "google/gemini-3-5-flash-lite",
          "wire_id": "gemini-3.5-flash-lite",
          "endpoint": "generate-content",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 2.5,
            "cache_read": 0.03,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/pricing",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 65536,
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Default minimal; docs recommend medium/high for multi-step subagent work — minimal can terminate prematurely on complex tasks. thinkingLevel replaces thinkingBudget; temperature/top_p/top_k deprecated on 3.6/3.7.",
            "returns": "thought_parts",
            "must_round_trip": "thought_signature",
            "incompatible_with": [],
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/thinking",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "generationConfig.thinkingConfig.thinkingLevel",
              "values": [
                "minimal",
                "low",
                "medium",
                "high"
              ],
              "default": "minimal"
            }
          }
        },
        {
          "model": "google/gemini-3-6-flash",
          "wire_id": "gemini-3.6-flash",
          "endpoint": "generate-content",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.75,
            "output": 3.75,
            "cache_read": 0.075,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/pricing",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 65536,
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Intro pricing (50% off) through 2026-12-31; thinkingLevel replaces thinkingBudget; temperature/top_p/top_k deprecated on 3.6/3.7.",
            "returns": "thought_parts",
            "must_round_trip": "thought_signature",
            "incompatible_with": [],
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/thinking",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "generationConfig.thinkingConfig.thinkingLevel",
              "values": [
                "minimal",
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-7-flash",
          "wire_id": "gemini-3.7-flash",
          "endpoint": "generate-content",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.75,
            "output": 3.75,
            "cache_read": 0.075,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/pricing",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 65536,
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "minimal unsupported (errors); thinkingLevel replaces thinkingBudget; temperature/top_p/top_k deprecated on 3.6/3.7.",
            "returns": "thought_parts",
            "must_round_trip": "thought_signature",
            "incompatible_with": [],
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/thinking",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "generationConfig.thinkingConfig.thinkingLevel",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-flash",
          "wire_id": "gemini-3-flash-preview",
          "endpoint": "generate-content",
          "status": "preview",
          "status_date": "",
          "cost": {
            "input": 0.5,
            "output": 3,
            "cache_read": 0.05,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/pricing",
              "verified": "2026-08-21"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 65536,
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/models",
              "verified": "2026-08-21"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "thinkingLevel replaces thinkingBudget on Gemini 3; cannot disable thinking.",
            "returns": "thought_parts",
            "must_round_trip": "thought_signature",
            "incompatible_with": [],
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/thinking",
              "verified": "2026-08-21"
            },
            "effort": {
              "param": "generationConfig.thinkingConfig.thinkingLevel",
              "values": [
                "minimal",
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "google/gemini-3-pro",
          "wire_id": "gemini-3-pro-preview",
          "endpoint": "generate-content",
          "status": "preview",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 12,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "notes": "Tiered: >200k doubles. Gemini 3 Pro text pricing now carried by gemini-3.1-pro-preview.",
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 65536,
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "thinkingLevel replaces thinkingBudget on Gemini 3; specifying both errors. Cannot disable thinking.",
            "returns": "thought_parts",
            "must_round_trip": "thought_signature",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.cloud.google.com/vertex-ai/generative-ai/docs/thinking",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "generationConfig.thinkingConfig.thinkingLevel",
              "values": [
                "low",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "google/gemma-4",
          "wire_id": "gemma-4-31b-it",
          "endpoint": "generate-content",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://deepmind.google/models/gemma",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Gemma 4 has configurable thinking modes (think token in the chat template); the Gemini API control parameter is not documented — verify. Serves the 31B dense variant. Free tier only; no paid tier.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://deepmind.google/models/gemma",
              "verified": "2026-08-20"
            }
          }
        }
      ]
    },
    {
      "id": "google-vertex",
      "name": "Google Vertex AI",
      "kind": "cloud_hosted",
      "urls": {
        "docs": "https://docs.cloud.google.com/vertex-ai",
        "console": "https://console.cloud.google.com"
      },
      "auth": [
        {
          "id": "adc",
          "type": "adc",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "GOOGLE_APPLICATION_CREDENTIALS"
          ],
          "extra_headers": {},
          "getting_credentials": "gcloud auth application-default login; classic endpoints reject API keys (express mode adds API-key auth for some endpoints).",
          "docs": "https://ai.google.dev/gemini-api/docs/migrate-to-cloud"
        }
      ],
      "endpoints": [
        {
          "id": "generate-content",
          "base_url": "https://{region}-aiplatform.googleapis.com",
          "path": "/v1/projects/{project}/locations/{region}/publishers/google/models/{model}:generateContent",
          "protocol": "google-generate-content"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "embeddings",
        "batch",
        "count_tokens",
        "prompt_caching"
      ],
      "quirks": [],
      "offerings": [
        {
          "model": "google/gemini-3-5-flash-lite",
          "wire_id": "gemini-3.5-flash-lite",
          "endpoint": "generate-content",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 2.5,
            "cache_read": 0.03,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud.google.com/vertex-ai/generative-ai/pricing",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 65536,
            "source": {
              "url": "https://cloud.google.com/vertex-ai/generative-ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Clone of the Gemini API offering; Vertex id — verify. Default minimal; docs recommend medium/high for multi-step subagent work — minimal can terminate prematurely on complex tasks. thinkingLevel replaces thinkingBudget; temperature/top_p/top_k deprecated on 3.6/3.7.",
            "returns": "thought_parts",
            "must_round_trip": "thought_signature",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.cloud.google.com/vertex-ai/generative-ai/docs/thinking",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "generationConfig.thinkingConfig.thinkingLevel",
              "values": [
                "minimal",
                "low",
                "medium",
                "high"
              ],
              "default": "minimal"
            }
          }
        },
        {
          "model": "google/gemini-3-6-flash",
          "wire_id": "gemini-3.6-flash",
          "endpoint": "generate-content",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.75,
            "output": 3.75,
            "cache_read": 0.075,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud.google.com/vertex-ai/generative-ai/pricing",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 65536,
            "source": {
              "url": "https://cloud.google.com/vertex-ai/generative-ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Clone of the Gemini API offering; Vertex id — verify. Intro pricing (50% off) through 2026-12-31; thinkingLevel replaces thinkingBudget; temperature/top_p/top_k deprecated on 3.6/3.7.",
            "returns": "thought_parts",
            "must_round_trip": "thought_signature",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.cloud.google.com/vertex-ai/generative-ai/docs/thinking",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "generationConfig.thinkingConfig.thinkingLevel",
              "values": [
                "minimal",
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-7-flash",
          "wire_id": "gemini-3.7-flash",
          "endpoint": "generate-content",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.75,
            "output": 3.75,
            "cache_read": 0.075,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud.google.com/vertex-ai/generative-ai/pricing",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 65536,
            "source": {
              "url": "https://cloud.google.com/vertex-ai/generative-ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Clone of the Gemini API offering; Vertex id — verify. minimal unsupported (errors); thinkingLevel replaces thinkingBudget; temperature/top_p/top_k deprecated on 3.6/3.7.",
            "returns": "thought_parts",
            "must_round_trip": "thought_signature",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.cloud.google.com/vertex-ai/generative-ai/docs/thinking",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "generationConfig.thinkingConfig.thinkingLevel",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-pro",
          "wire_id": "gemini-3-pro-preview",
          "endpoint": "generate-content",
          "status": "preview",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 12,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "notes": "Tiered: >200k doubles. Gemini 3 Pro text pricing now carried by gemini-3.1-pro-preview.",
            "source": {
              "url": "https://ai.google.dev/gemini-api/docs/pricing",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "thinkingLevel clone of the Gemini API offering; Vertex id — verify. Cannot disable thinking; thinkingLevel replaces thinkingBudget on Gemini 3.",
            "returns": "thought_parts",
            "must_round_trip": "thought_signature",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.cloud.google.com/vertex-ai/generative-ai/docs/thinking",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "generationConfig.thinkingConfig.thinkingLevel",
              "values": [
                "low",
                "high"
              ],
              "default": "high"
            }
          }
        }
      ]
    },
    {
      "id": "hetzner",
      "name": "Hetzner Inference (Experiments)",
      "kind": "first_party",
      "urls": {
        "docs": "https://docs.hetzner.com/general/company-and-policy/experiments/inference/",
        "console": "https://experiments.hetzner.com"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "HETZNER_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create a token in the Experiments dashboard; free while experimental, no SLA.",
          "docs": "https://docs.hetzner.com/general/company-and-policy/experiments/inference/"
        }
      ],
      "endpoints": [
        {
          "id": "chat",
          "base_url": "https://inference.hetzner.com",
          "path": "/api/v1/chat/completions",
          "protocol": "openai-chat"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Experimental 'as is' program (launched ~2026-07): free, EU-based, no production guarantees; email notice before changes. GET /api/v1/models is definitive for the lineup.",
          "docs": "https://docs.hetzner.com/general/company-and-policy/experiments/inference/"
        },
        {
          "text": "Limits per key: 10 requests/60s, 4M input and 100k output tokens/60s. Reasoning control is undocumented; a community-reported chat_template_kwargs.enable_thinking flag exists for Qwen3.6.",
          "docs": "https://docs.hetzner.com/general/company-and-policy/experiments/inference/"
        }
      ],
      "offerings": [
        {
          "model": "alibaba/qwen3-8-27b",
          "wire_id": "Qwen/Qwen3.8-27B",
          "endpoint": "chat",
          "status": "preview",
          "status_date": "",
          "cost": {
            "input": 0,
            "output": 0,
            "cache_read": null,
            "cache_write": null,
            "free": true,
            "source": {
              "url": "https://docs.hetzner.com/general/company-and-policy/experiments/inference/",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://inference.hetzner.com/api/v1/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning control is undocumented; chat_template_kwargs.enable_thinking is community-reported for Qwen3.6 — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.hetzner.com/general/company-and-policy/experiments/inference/",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "chat_template_kwargs.enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        }
      ]
    },
    {
      "id": "io-intelligence",
      "name": "IO Intelligence (io.net)",
      "kind": "first_party",
      "urls": {
        "docs": "https://io.net/docs/guides/intelligence/io-intelligence-apis",
        "console": "https://ai.io.net",
        "pricing": "https://io.net/docs/guides/payment/io-intelligence-payments"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "IOINTELLIGENCE_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create keys at ai.io.net/ai/api-keys with expiry and permission scoping.",
          "docs": "https://io.net/docs/guides/intelligence/api-keys-and-secrets"
        }
      ],
      "endpoints": [
        {
          "id": "chat",
          "base_url": "https://api.intelligence.io.solutions",
          "path": "/api/v1/chat/completions",
          "protocol": "openai-chat"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Base path is /api/v1 (not /v1) on the io.solutions domain; api.intelligence.io.net does not resolve.",
          "docs": "https://io.net/docs/guides/intelligence/io-intelligence-apis"
        },
        {
          "text": "Per-token prices (USD, scientific notation) come from GET /api/v1/models; model access gated by min_access_tier 1-3 by plan (Standard free-light/Professional $15/Developer $150 daily credits).",
          "docs": "https://api.intelligence.io.solutions/api/v1/models"
        },
        {
          "text": "reasoning.effort none/low/medium/high is primary; legacy top-level reasoning_effort still accepted; MiniMax-M2.7 is always-on (none is a no-op; some models 400 on none).",
          "docs": "https://io.net/docs/guides/intelligence/reasoning-for-chat-completions"
        }
      ],
      "offerings": [
        {
          "model": "alibaba/qwen3-8-27b",
          "wire_id": "Qwen/Qwen3.8-27B",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.45,
            "output": 3.25,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 65500,
            "output": 65536,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-flash",
          "wire_id": "deepseek-ai/DeepSeek-V4-Flash",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.201,
            "output": 0.52,
            "cache_read": 0.1,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 32768,
            "output": 32768,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-pro",
          "wire_id": "deepseek-ai/DeepSeek-V4-Pro",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.4872,
            "output": 3.0244,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://io.net/docs/guides/intelligence/reasoning-for-chat-completions",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "google/gemma-4",
          "wire_id": "google/gemma-4-26b-a4b-it",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.11,
            "output": 0.408,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262142,
            "output": null,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Gemma 4 thinking is configurable upstream (enable_thinking, default on); io.net lists reasoning support without a documented control — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "meta/llama-3-3-70b",
          "wire_id": "meta-llama/Llama-3.3-70B-Instruct",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.638,
            "output": 0.768,
            "cache_read": 0.319,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "none",
            "mandatory": false,
            "default": "off",
            "notes": "Llama 3.3 is not a reasoning model; no reasoning parameter on this surface.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://io.net/docs/guides/intelligence/reasoning-for-chat-completions",
              "verified": "2026-08-19"
            }
          }
        },
        {
          "model": "meta/llama-4-maverick",
          "wire_id": "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.274,
            "output": 0.8992,
            "cache_read": 0.137,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "none",
            "mandatory": false,
            "default": "off",
            "notes": "Llama 4 is not a reasoning model; no reasoning parameter on this surface.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://io.net/docs/guides/intelligence/reasoning-for-chat-completions",
              "verified": "2026-08-19"
            }
          }
        },
        {
          "model": "minimax/minimax-m2-5",
          "wire_id": "MiniMaxAI/MiniMax-M2.5",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.348,
            "output": 1.366,
            "cache_read": 0.174,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 196600,
            "output": null,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "MiniMax M-series cannot disable thinking upstream.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "minimax/minimax-m2-7",
          "wire_id": "MiniMaxAI/MiniMax-M2.7",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.404,
            "output": 1.496,
            "cache_read": 0.202,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262100,
            "output": 65536,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "MiniMax M-series cannot disable thinking upstream.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-5",
          "wire_id": "moonshotai/Kimi-K2.5",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.544,
            "output": 2.855,
            "cache_read": 0.272,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-6",
          "wire_id": "moonshotai/Kimi-K2.6",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.685,
            "output": 3.312,
            "cache_read": 0.343,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262142,
            "output": null,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-7-code",
          "wire_id": "moonshotai/Kimi-K2.7-Code",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.008,
            "output": 4.51,
            "cache_read": 0.504,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Kimi K2.7 Code toggles thinking enabled-only upstream (off sends empty string).",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "moonshot/kimi-k3",
          "wire_id": "moonshotai/Kimi-K3",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 3,
            "output": 15,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://io.net/docs/guides/intelligence/reasoning-for-chat-completions",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "openai/gpt-oss-120b",
          "wire_id": "openai/gpt-oss-120b",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.188,
            "output": 0.7,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://io.net/docs/guides/intelligence/reasoning-for-chat-completions",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "none suppresses reasoning output; the model still reasons internally."
            }
          }
        },
        {
          "model": "openai/gpt-oss-20b",
          "wire_id": "openai/gpt-oss-20b",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.067,
            "output": 0.24,
            "cache_read": 0.034,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 64000,
            "output": null,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "gpt-oss cannot disable thinking upstream.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "xiaomi/mimo-v2-5",
          "wire_id": "XiaomiMiMo/MiMo-V2.5",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.193,
            "output": 0.627,
            "cache_read": 0.097,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "zai/glm-4-6",
          "wire_id": "zai-org/GLM-4.6",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.536,
            "output": 2.07,
            "cache_read": 0.268,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 131072,
            "output": null,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "zai/glm-4-7",
          "wire_id": "zai-org/GLM-4.7",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.88,
            "output": 2.37,
            "cache_read": 0.44,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 202752,
            "output": null,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "zai/glm-4-7-flash",
          "wire_id": "zai-org/GLM-4.7-Flash",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.0626,
            "output": 0.4,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 200000,
            "output": null,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "zai/glm-5-1",
          "wire_id": "zai-org/GLM-5.1",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.314,
            "output": 4.268,
            "cache_read": 0.657,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 202750,
            "output": 32768,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "zai/glm-5-2",
          "wire_id": "zai-org/GLM-5.2",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.552,
            "output": 4.884,
            "cache_read": 0.776,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": 131072,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "zai/glm-5-3",
          "wire_id": "zai-org/GLM-5.3",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.4,
            "output": 4.4,
            "cache_read": 0.7,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 131072,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "GLM-5.3 thinks always; thinking.type disabled errors upstream — verify gateway behavior.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "zai/glm-5-3-flash",
          "wire_id": "zai-org/GLM-5.3-Flash",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.15,
            "output": 0.5,
            "cache_read": 0.075,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": 131072,
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "GLM-5.3 Flash thinks always; enabled-only upstream — verify gateway behavior.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.intelligence.io.solutions/api/v1/models",
              "verified": "2026-08-28"
            }
          }
        }
      ]
    },
    {
      "id": "kimi-coding",
      "name": "Kimi for Coding",
      "kind": "subscription",
      "urls": {
        "docs": "https://www.kimi.com/code/docs/",
        "console": "https://www.kimi.com/code/console"
      },
      "auth": [
        {
          "id": "plan-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [],
          "extra_headers": {},
          "getting_credentials": "Create keys in the Kimi Code Console (shown once). For Claude Code: ANTHROPIC_BASE_URL=https://api.kimi.com/coding/ and ANTHROPIC_API_KEY=<coding key>.",
          "docs": "https://www.kimi.com/code/docs/en/third-party-tools/claude-code.html"
        }
      ],
      "endpoints": [
        {
          "id": "coding-openai",
          "base_url": "https://api.kimi.com",
          "path": "/coding/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "coding-anthropic",
          "base_url": "https://api.kimi.com",
          "path": "/coding",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Model ids on the coding endpoints: kimi-for-coding, kimi-for-coding-highspeed, k3 (1M ctx, ~2x quota), k3-256k. The [1m] suffix (kimi-k3[1m]) is a Claude Code env convention only — raw API uses k3/kimi-k3. Claude Code /effort maps to k3 low/high/max.",
          "docs": "https://www.kimi.com/code/docs/en/third-party-tools/claude-code.html"
        },
        {
          "text": "kimi-for-coding is the K2.7-code-class model; highspeed variant doubles speed and quota consumption.",
          "docs": "https://www.kimi.com/code/docs/kimi-code/faq.html"
        }
      ],
      "plan": {
        "price_usd": 19,
        "period": "monthly",
        "quota": "Kimi Code has its own rolling 5-hour + weekly limits plus shared agent credits (60/150/360/720 by tier); concurrency-limited",
        "notes": "Included with Kimi membership — tiers Moderato $19, Allegretto $39, Allegro $99, Vivace $199 (annual discounts); K3 access from Moderato up. Coding keys are distinct from Open Platform keys and never interchangeable (top 401 cause).",
        "docs": "https://www.kimi.ai/help/membership/membership-pricing"
      },
      "offerings": [
        {
          "model": "moonshot/kimi-k2-7-code",
          "wire_id": "kimi-for-coding-highspeed",
          "endpoint": "coding-openai",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://www.kimi.com/code/docs/en/third-party-tools/claude-code.html",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": ""
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-7-code",
          "wire_id": "kimi-for-coding",
          "endpoint": "coding-openai",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://www.kimi.com/code/docs/en/third-party-tools/claude-code.html",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": ""
            }
          }
        },
        {
          "model": "moonshot/kimi-k3",
          "wire_id": "k3-256k",
          "endpoint": "coding-anthropic",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "262144-context variant; control effort via Claude Code /effort (low/high/max) — raw wire param on this surface not documented; verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://www.kimi.com/code/docs/en/third-party-tools/claude-code.html",
              "verified": "2026-08-19"
            }
          }
        },
        {
          "model": "moonshot/kimi-k3",
          "wire_id": "k3",
          "endpoint": "coding-openai",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://www.kimi.com/code/docs/en/third-party-tools/claude-code.html",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "high",
                "max"
              ],
              "default": "max"
            }
          }
        }
      ]
    },
    {
      "id": "meta",
      "name": "Meta Model API",
      "kind": "first_party",
      "urls": {
        "docs": "https://dev.meta.ai/docs/overview/",
        "console": "https://dev.meta.ai",
        "pricing": "https://dev.meta.ai/docs/pricing-rate-limits/"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "MODEL_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create keys in the dev.meta.ai dashboard (API Keys tab). For Claude Code: ANTHROPIC_BASE_URL=https://api.meta.ai with ANTHROPIC_AUTH_TOKEN=<key>.",
          "docs": "https://dev.meta.ai/docs/quickstart/"
        }
      ],
      "endpoints": [
        {
          "id": "v1-responses",
          "base_url": "https://api.meta.ai",
          "path": "/v1/responses",
          "protocol": "openai-responses"
        },
        {
          "id": "v1-chat-completions",
          "base_url": "https://api.meta.ai",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "v1-messages",
          "base_url": "https://api.meta.ai",
          "path": "/v1/messages",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "files"
      ],
      "quirks": [
        {
          "text": "The Llama API (api.llama.com/compat/v1) was retired 2026-07-06 — there is no first-party hosted Llama inference; Llama models are open weights served by third parties.",
          "docs": "https://www.promptfoo.dev/docs/providers/llamaApi/"
        },
        {
          "text": "reasoning_effort none returns 400 (always-on reasoning). Chat Completions exposes reasoning_content redacted to empty for external callers; use the Responses surface for summaries and encrypted reasoning replay. No logprobs, no n>1, no stop.",
          "docs": "https://dev.meta.ai/docs/reasoning"
        },
        {
          "text": "All keys share team limits: Standard 3,000 RPM / 4M TPM. Contributor tier (muse-spark-1.2-contributor) trades a heavy discount for prompts possibly training future Meta models.",
          "docs": "https://dev.meta.ai/docs/pricing-rate-limits/"
        }
      ],
      "offerings": [
        {
          "model": "meta/muse-spark-1-2",
          "wire_id": "muse-spark-1.2",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 4.25,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://dev.meta.ai/docs/pricing-rate-limits/",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 131072,
            "source": {
              "url": "https://dev.meta.ai/docs/pricing-rate-limits/",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "reasoning_content is redacted to empty for external callers on this surface.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://dev.meta.ai/docs/reasoning",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "medium",
              "notes": "Default effort is not documented — verify."
            }
          }
        },
        {
          "model": "meta/muse-spark-1-2",
          "wire_id": "muse-spark-1.2-contributor",
          "endpoint": "v1-responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.1,
            "output": 0.2,
            "cache_read": 0.002,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://dev.meta.ai/docs/pricing-rate-limits/",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 131072,
            "source": {
              "url": "https://dev.meta.ai/docs/pricing-rate-limits/",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Contributor tier: heavy discount in exchange for prompts possibly training future Meta models.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [],
            "source": {
              "url": "https://dev.meta.ai/docs/reasoning",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "medium",
              "notes": "Default effort is not documented — verify."
            }
          }
        },
        {
          "model": "meta/muse-spark-1-2",
          "wire_id": "muse-spark-1.2",
          "endpoint": "v1-responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 4.25,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://dev.meta.ai/docs/pricing-rate-limits/",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 131072,
            "source": {
              "url": "https://dev.meta.ai/docs/pricing-rate-limits/",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [],
            "source": {
              "url": "https://dev.meta.ai/docs/reasoning",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "medium",
              "notes": "Default effort is not documented — verify."
            }
          }
        }
      ]
    },
    {
      "id": "minimax",
      "name": "MiniMax",
      "kind": "first_party",
      "urls": {
        "docs": "https://platform.minimax.io/docs/api-reference/api-overview",
        "console": "https://platform.minimax.io"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "MINIMAX_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "platform.minimax.io → API Keys. Subscription Keys are a separate credential type issued under Billing → Token Plan.",
          "docs": "https://platform.minimax.io/docs/api-reference/api-overview"
        }
      ],
      "endpoints": [
        {
          "id": "chat-completions",
          "base_url": "https://api.minimax.io",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "anthropic",
          "base_url": "https://api.minimax.io",
          "path": "/anthropic",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "reasoning_split: true splits thinking into message.reasoning_content; without it thinking arrives inline as <think>...</think> inside content. China base: api.minimaxi.com.",
          "docs": "https://platform.minimax.io/docs/api-reference/text-chat-openai"
        }
      ],
      "offerings": [
        {
          "model": "minimax/minimax-m2-5",
          "wire_id": "MiniMax-M2.5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 1.2,
            "cache_read": 0.03,
            "cache_write": 0.375,
            "free": false,
            "source": {
              "url": "https://platform.minimax.io/docs/guides/pricing-paygo",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 204800,
            "output": null,
            "source": {
              "url": "https://platform.minimax.io/docs/guides/text-generation",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "thinking: {type: \"disabled\"} has no effect on M2.x — thinking cannot be turned off.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://platform.minimaxi.com/docs/api-reference/text-chat-openai",
              "verified": "2026-08-18"
            }
          }
        },
        {
          "model": "minimax/minimax-m3",
          "wire_id": "MiniMax-M3",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 1.2,
            "cache_read": 0.06,
            "cache_write": null,
            "free": false,
            "notes": "50% off ≤512k promo made permanent; >512k doubles; Priority tier 1.5x.",
            "source": {
              "url": "https://platform.minimax.io/docs/guides/pricing-paygo",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://www.minimax.io/blog/minimax-m3",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "adaptive",
            "notes": "thinking.type is adaptive|disabled (no 'enabled'); adaptive lets the model decide.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://platform.minimaxi.com/docs/api-reference/text-chat-openai",
              "verified": "2026-08-18"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "adaptive",
              "off": "disabled"
            }
          }
        }
      ]
    },
    {
      "id": "minimax-token-plan",
      "name": "MiniMax Token Plan",
      "kind": "subscription",
      "urls": {
        "docs": "https://platform.minimax.io/docs/token-plan/intro",
        "console": "https://platform.minimax.io/subscribe/token-plan"
      },
      "auth": [
        {
          "id": "subscription-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "MINIMAX_SUBSCRIPTION_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Subscribe at platform.minimax.io/subscribe/token-plan; the Subscription Key is issued separately from standard API keys under Billing > Token Plan.",
          "docs": "https://platform.minimax.io/docs/token-plan/intro"
        }
      ],
      "endpoints": [
        {
          "id": "anthropic",
          "base_url": "https://api.minimax.io",
          "path": "/anthropic",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Anthropic-compatible endpoint for coding tools (Claude Code). China base: https://api.minimaxi.com/anthropic.",
          "docs": "https://platform.minimax.io/docs/token-plan/claude-code"
        }
      ],
      "plan": {
        "price_usd": 20,
        "period": "monthly",
        "quota": "Plus ~1.7B tokens/mo (3-4 concurrent agents); Max ~5.1B; higher tiers to ~$120",
        "notes": "Subscription Key is a separate credential type from pay-as-you-go API keys, issued under Billing > Token Plan. Sources conflict on Plus pricing ($20 vs $40) — verify at the subscribe page. For Claude Code: ANTHROPIC_BASE_URL=https://api.minimax.io/anthropic + ANTHROPIC_AUTH_TOKEN=<Subscription Key>.",
        "docs": "https://platform.minimax.io/docs/token-plan/claude-code"
      },
      "offerings": [
        {
          "model": "minimax/minimax-m2-5",
          "wire_id": "MiniMax-M2.5",
          "endpoint": "anthropic",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "M2.x thinking cannot be turned off on any surface; thinking {type: disabled} has no effect.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://platform.minimaxi.com/docs/api-reference/text-chat-openai",
              "verified": "2026-08-18"
            }
          }
        },
        {
          "model": "minimax/minimax-m3",
          "wire_id": "MiniMax-M3",
          "endpoint": "anthropic",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "adaptive",
            "notes": "thinking.type is adaptive|disabled (no enabled). Verified on the OpenAI-compatible chat surface; exact wire shape on the /anthropic surface — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://platform.minimaxi.com/docs/api-reference/text-chat-openai",
              "verified": "2026-08-18"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "adaptive",
              "off": "disabled"
            }
          }
        }
      ]
    },
    {
      "id": "mistral",
      "name": "Mistral",
      "kind": "first_party",
      "urls": {
        "docs": "https://docs.mistral.ai/api/endpoint/chat",
        "console": "https://console.mistral.ai"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "MISTRAL_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create a key at console.mistral.ai (La Plateforme).",
          "docs": "https://docs.mistral.ai/api/endpoint/chat"
        }
      ],
      "endpoints": [
        {
          "id": "v1-chat-completions",
          "base_url": "https://api.mistral.ai",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "embeddings",
        "files",
        "batch"
      ],
      "quirks": [
        {
          "text": "With reasoning_effort high, message.content becomes a list of chunks: ThinkChunk (type thinking) followed by TextChunk; replay the full assistant message including ThinkChunk in multi-turn or quality degrades. No Responses API. Magistral reasoning models deprecated; reasoning_effort high/none supported on mistral-small-latest and mistral-medium-3-5.",
          "docs": "https://docs.mistral.ai/capabilities/reasoning/"
        }
      ],
      "offerings": [
        {
          "model": "mistral/mistral-medium-3-5",
          "wire_id": "mistral-medium-3.5",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.5,
            "output": 7.5,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://mistral.ai/pricing/api/",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 256000,
            "output": null,
            "source": {
              "url": "https://docs.mistral.ai/models/",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Default effort not documented; verify at docs.mistral.ai/capabilities/reasoning. With high, content is ThinkChunk + TextChunk blocks; replay ThinkChunk in multi-turn.",
            "returns": "thinking_blocks",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.mistral.ai/capabilities/reasoning/",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "high"
              ],
              "default": "none"
            }
          }
        },
        {
          "model": "mistral/mistral-small-latest",
          "wire_id": "mistral-small-latest",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.15,
            "output": 0.6,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://mistral.ai/pricing/api/",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 256000,
            "output": null,
            "source": {
              "url": "https://docs.mistral.ai/models/",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Default effort not documented; verify at docs.mistral.ai/capabilities/reasoning. With high, content is ThinkChunk + TextChunk blocks; replay ThinkChunk in multi-turn.",
            "returns": "thinking_blocks",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.mistral.ai/capabilities/reasoning/",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "high"
              ],
              "default": "none"
            }
          }
        }
      ]
    },
    {
      "id": "moonshot",
      "name": "Moonshot Kimi",
      "kind": "first_party",
      "urls": {
        "docs": "https://platform.kimi.ai/docs/overview",
        "console": "https://platform.kimi.ai/console/api-keys",
        "pricing": "https://platform.kimi.ai/docs/models"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "MOONSHOT_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create keys at the Kimi Open Platform (intl) or platform.kimi.com (CN, Chinese phone required). Keys are region-bound: .ai and .cn are not interchangeable.",
          "docs": "https://platform.kimi.ai/docs/overview"
        }
      ],
      "endpoints": [
        {
          "id": "chat-completions",
          "base_url": "https://api.moonshot.ai",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "anthropic",
          "base_url": "https://api.moonshot.ai",
          "path": "/anthropic",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "files",
        "batch"
      ],
      "quirks": [
        {
          "text": "China base: https://api.moonshot.cn/v1 (console platform.kimi.com). No embeddings endpoint. Partial Mode (partial: true) and ms:// file references are Moonshot extensions. Context caching is automatic for prompts over 256 tokens. Rate limits tier by cumulative top-up.",
          "docs": "https://platform.kimi.ai/docs/overview"
        },
        {
          "text": "Sampling params are fixed on k-series models (temperature 1.0, top_p 0.95; k2.6 uses 0.6 outside thinking mode) and error if set otherwise; only moonshot-v1 allows modifying temperature. tool_choice required is k3-only.",
          "docs": "https://platform.kimi.ai/docs/guide/use-thinking-models"
        },
        {
          "text": "The /anthropic endpoint (for Claude Code et al.) uses OPEN PLATFORM keys — never Kimi Code keys. WebFetch is unsupported there; WebSearch 400s on kimi-k2.7-code with thinking off.",
          "docs": "https://platform.kimi.ai/docs/guide/claude-code-kimi"
        }
      ],
      "offerings": [
        {
          "model": "moonshot/kimi-k2-5",
          "wire_id": "kimi-k2.5",
          "endpoint": "chat-completions",
          "status": "deprecated",
          "status_date": "2026-08-31",
          "cost": {
            "input": 0.6,
            "output": 3,
            "cache_read": 0.1,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://platform.kimi.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://platform.kimi.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Default thinking state not documented; verify. k2.5 has no keep param — round-trip rule not documented.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://platform.kimi.ai/docs/models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-6",
          "wire_id": "kimi-k2.6",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.95,
            "output": 4,
            "cache_read": 0.16,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://platform.kimi.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://platform.kimi.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "thinking.keep: null (default, drops reasoning history) or all (Preserved Thinking).",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [
              "temperature",
              "top_p",
              "n",
              "presence_penalty",
              "frequency_penalty"
            ],
            "source": {
              "url": "https://platform.kimi.ai/docs/guide/use-thinking-models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-7-code",
          "wire_id": "kimi-k2.7-code",
          "endpoint": "anthropic",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.95,
            "output": 4,
            "cache_read": 0.19,
            "cache_write": null,
            "free": false,
            "notes": "Same as native rates (Claude Code guide links the native pricing page).",
            "source": {
              "url": "https://platform.kimi.ai/docs/pricing/chat-k27-code",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": true,
            "default": "on",
            "notes": "Anthropic-compatible surface (for Claude Code et al.); thinking.type accepts only enabled.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://platform.kimi.ai/docs/guide/claude-code-kimi",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": ""
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-7-code",
          "wire_id": "kimi-k2.7-code-highspeed",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.9,
            "output": 8,
            "cache_read": 0.38,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://platform.kimi.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://platform.kimi.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": true,
            "default": "on",
            "notes": "~180-260 tok/s; exactly 2x pricing.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [
              "temperature",
              "top_p",
              "n",
              "presence_penalty",
              "frequency_penalty"
            ],
            "source": {
              "url": "https://platform.kimi.ai/docs/guide/use-thinking-models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": ""
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-7-code",
          "wire_id": "kimi-k2.7-code",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.95,
            "output": 4,
            "cache_read": 0.19,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://platform.kimi.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://platform.kimi.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": true,
            "default": "on",
            "notes": "thinking.type accepts ONLY enabled — disabled errors; thinking.keep treated as all.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [
              "temperature",
              "top_p",
              "n",
              "presence_penalty",
              "frequency_penalty"
            ],
            "source": {
              "url": "https://platform.kimi.ai/docs/guide/use-thinking-models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": ""
            }
          }
        },
        {
          "model": "moonshot/kimi-k3",
          "wire_id": "kimi-k3",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 3,
            "output": 15,
            "cache_read": 0.3,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://platform.kimi.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://platform.kimi.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Always-on reasoning; do NOT pass thinking on k3. Preserved Thinking always on. Claude Code /effort low/medium/high/xhigh/max maps to low/high/max.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [
              "temperature",
              "top_p",
              "n",
              "presence_penalty",
              "frequency_penalty"
            ],
            "source": {
              "url": "https://platform.kimi.ai/docs/guide/use-reasoning-effort",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "high",
                "max"
              ],
              "default": "max"
            }
          }
        }
      ]
    },
    {
      "id": "near-ai",
      "name": "NEAR AI Cloud",
      "kind": "first_party",
      "urls": {
        "docs": "https://docs.near.ai/cloud/quickstart",
        "console": "https://cloud.near.ai"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "NEAR_AI_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Keys from cloud.near.ai Developer Dashboard; prepaid Usage Credits (or NEAR staking yield). The old api.near.ai went offline 2025-10-31.",
          "docs": "https://docs.near.ai/cloud/quickstart"
        }
      ],
      "endpoints": [
        {
          "id": "chat",
          "base_url": "https://cloud-api.near.ai",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "responses",
          "base_url": "https://cloud-api.near.ai",
          "path": "/v1/responses",
          "protocol": "openai-responses"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "embeddings",
        "image_gen"
      ],
      "quirks": [
        {
          "text": "Inference runs in hardware TEEs with attestation (/v1/attestation/report) and response signatures; direct per-model endpoints at {slug}.completions.near.ai terminate TLS inside the enclave.",
          "docs": "https://docs.near.ai/cloud/quickstart"
        },
        {
          "text": "Per-model prices live in GET /v1/models (public, no auth); frontier OpenAI/Anthropic/Google models are resold through the gateway in OpenAI format.",
          "docs": "https://cloud-api.near.ai/v1/models"
        },
        {
          "text": "Reasoning params vary by model: reasoning_effort (gpt-oss, always-on), chat_template_kwargs.enable_thinking (GLM/Qwen), chat_template_kwargs.thinking (Kimi — not enable_thinking).",
          "docs": "https://docs.near.ai/cloud/reasoning-models/"
        },
        {
          "text": "The OpenAI Responses-compatible /v1/responses endpoint on the gateway is in beta.",
          "docs": "https://docs.near.ai/cloud/guides/openai-compatibility"
        }
      ],
      "offerings": [
        {
          "model": "alibaba/qwen3-7-max",
          "wire_id": "qwen/qwen3.7-max",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2.8,
            "output": 7.5,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 16384,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": false
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Thinking mode exists upstream (Alibaba Model Studio); NEAR does not list reasoning in supported features and the gateway control is undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "alibaba/qwen3-8-27b",
          "wire_id": "Qwen/Qwen3.8-27B",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.44,
            "output": 3.3,
            "cache_read": 0.044,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": 8192,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "NEAR documents chat_template_kwargs.enable_thinking for Qwen.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-28"
            },
            "toggle": {
              "param": "chat_template_kwargs.enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "anthropic/claude-fable-5",
          "wire_id": "anthropic/claude-fable-5",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 10,
            "output": 50,
            "cache_read": 1,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 128000,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": true,
            "default": "on",
            "notes": "Claude 5-series adaptive effort upstream (output_config.effort); fable-5 cannot disable thinking; gateway control undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "anthropic/claude-fable-5-1",
          "wire_id": "anthropic/claude-fable-5-1",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 10,
            "output": 50,
            "cache_read": 0.25,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-09-04"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 128000,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": true,
            "default": "on",
            "notes": "Fable line cannot disable thinking; NEAR gateway control undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-09-04"
            }
          }
        },
        {
          "model": "anthropic/claude-haiku-4-5",
          "wire_id": "anthropic/claude-haiku-4-5",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1,
            "output": 5,
            "cache_read": 0.1,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 200000,
            "output": 64000,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Claude adaptive effort upstream; gateway control undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-4-6",
          "wire_id": "anthropic/claude-opus-4-6",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 200000,
            "output": 32768,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Claude 4.x adaptive effort upstream; NEAR lists reasoning in supported features; gateway control undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-4-7",
          "wire_id": "anthropic/claude-opus-4-7",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 32768,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Claude 4.x adaptive effort upstream; NEAR lists reasoning in supported features; gateway control undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-4-8",
          "wire_id": "anthropic/claude-opus-4-8",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Claude 4.x adaptive effort upstream; NEAR lists reasoning in supported features; gateway control undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-5",
          "wire_id": "anthropic/claude-opus-5",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 128000,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Claude 5-series adaptive effort upstream (output_config.effort); gateway control undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "anthropic/claude-sonnet-5",
          "wire_id": "anthropic/claude-sonnet-5",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 10,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 128000,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Claude 5-series adaptive effort upstream (output_config.effort); gateway control undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-flash",
          "wire_id": "deepseek-ai/DeepSeek-V4-Flash",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.17,
            "output": 0.35,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Follows the upstream DeepSeek convention; not documented here — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "google/gemini-3-5-flash",
          "wire_id": "google/gemini-3.5-flash",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.5,
            "output": 9,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": 8192,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "NEAR does not list reasoning in supported features; upstream control is thinking_level (minimal/low/medium/high, default medium). No gateway control documented — verify before relying.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "google/gemini-3-8-flash",
          "wire_id": "google/gemini-3.8-flash",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.75,
            "output": 3.75,
            "cache_read": 0.075,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-09-04"
            }
          },
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": false,
            "structured_output": false,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "NEAR does not list reasoning in supported features; upstream control is thinking_level — verify gateway support.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-09-04"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemma-4",
          "wire_id": "google/gemma-4-31B-it",
          "endpoint": "chat",
          "status": "retired",
          "status_date": "2026-08-27",
          "cost": {
            "input": 0.13,
            "output": 0.4,
            "cache_read": 0.026,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 262144,
            "output": 8192,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning listed as a supported feature in NEAR's live model metadata; the control parameter for Gemma is not documented — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-20"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-6",
          "wire_id": "moonshotai/kimi-k2.6",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.81,
            "output": 3.85,
            "cache_read": 0.41,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": 8192,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "NEAR documents chat_template_kwargs.thinking for Kimi (not enable_thinking).",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-28"
            },
            "toggle": {
              "param": "chat_template_kwargs.thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "moonshot/kimi-k3",
          "wire_id": "moonshotai/kimi-k3",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 3.3,
            "output": 16.5,
            "cache_read": 0.33,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 65535,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "NEAR documents chat_template_kwargs.thinking for Kimi (not enable_thinking).",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-28"
            },
            "toggle": {
              "param": "chat_template_kwargs.thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "openai/gpt-5",
          "wire_id": "openai/gpt-5",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 10,
            "cache_read": 0.125,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 400000,
            "output": 16384,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-1",
          "wire_id": "openai/gpt-5.1",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 10,
            "cache_read": 0.125,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 400000,
            "output": 16384,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-4",
          "wire_id": "openai/gpt-5.4",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2.5,
            "output": 15,
            "cache_read": 0.25,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1050000,
            "output": 16384,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Default effort is none upstream.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "none"
            }
          }
        },
        {
          "model": "openai/gpt-5-4-mini",
          "wire_id": "openai/gpt-5.4-mini",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.75,
            "output": 4.5,
            "cache_read": 0.075,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 400000,
            "output": 16384,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Default effort is none upstream.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "none"
            }
          }
        },
        {
          "model": "openai/gpt-5-4-nano",
          "wire_id": "openai/gpt-5.4-nano",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.2,
            "output": 1.25,
            "cache_read": 0.02,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 400000,
            "output": 16384,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Default effort is none upstream.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "none"
            }
          }
        },
        {
          "model": "openai/gpt-5-5",
          "wire_id": "openai/gpt-5.5",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 30,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1050000,
            "output": 16384,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium",
              "notes": "Default not documented — verify."
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "openai/gpt-5.6-luna",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.2,
            "output": 1.2,
            "cache_read": 0.02,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-09-04"
            }
          },
          "limits": {
            "context": 1050000,
            "output": 128000,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "GPT-5.6 cost tier; reasoning.mode standard|pro upstream.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-09-04"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "openai/gpt-5.6-sol",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 4,
            "output": 20,
            "cache_read": 0.4,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1050000,
            "output": 128000,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "GPT-5.6 flagship tier; reasoning.mode standard|pro upstream.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-6",
          "wire_id": "openai/gpt-6-astra",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 10,
            "output": 50,
            "cache_read": 1,
            "cache_write": null,
            "free": false,
            "notes": "Short-context tier; over 272K tokens $20/$75 (OpenAI list; NEAR pricing shown).",
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-09-10"
            }
          },
          "limits": {
            "context": 1050000,
            "output": 128000,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-09-10"
            }
          },
          "features": {
            "streaming": true,
            "tools": false,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "NEAR lists reasoning in supported features; effort low-max with no none upstream, default undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-09-10"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-oss-120b",
          "wire_id": "openai/gpt-oss-120b",
          "endpoint": "chat",
          "status": "retired",
          "status_date": "2026-08-27",
          "cost": {
            "input": 0.15,
            "output": 0.55,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 131072,
            "output": null,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/o3",
          "wire_id": "openai/o3",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 8,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 200000,
            "output": 32768,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "zai/glm-5-2",
          "wire_id": "z-ai/glm-5.2",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.4,
            "output": 4.4,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-19"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Follows the GLM chat_template_kwargs.enable_thinking convention; not documented for GLM-5.2 specifically — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "chat_template_kwargs.enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "zai/glm-5-3-flash",
          "wire_id": "z-ai/glm-5.3-flash",
          "endpoint": "chat",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.15,
            "output": 0.5,
            "cache_read": 0.035,
            "cache_write": null,
            "free": false,
            "notes": "Near’s own cache-read price (0.035) sits just above Z.ai’s 0.03 list; input/output match the launch promo.",
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-09-10"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 131072,
            "source": {
              "url": "https://cloud-api.near.ai/v1/models",
              "verified": "2026-09-10"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "GLM-5.3 Flash thinks always (enabled-only upstream); NEAR gateway control undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.near.ai/cloud/reasoning-models/",
              "verified": "2026-09-10"
            }
          }
        }
      ]
    },
    {
      "id": "nvidia",
      "name": "NVIDIA NIM API",
      "kind": "first_party",
      "urls": {
        "docs": "https://docs.api.nvidia.com/nim/reference/llm-apis",
        "console": "https://build.nvidia.com"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "NVIDIA_API_KEY"
          ],
          "key_prefix": "nvapi-",
          "extra_headers": {},
          "getting_credentials": "Generate nvapi- keys at build.nvidia.com (free with the NVIDIA Developer Program).",
          "docs": "https://build.nvidia.com"
        }
      ],
      "endpoints": [
        {
          "id": "v1-chat-completions",
          "base_url": "https://integrate.api.nvidia.com",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "embeddings"
      ],
      "quirks": [
        {
          "text": "No pay-as-you-go: hosted access is free-tier rate limited (~40 RPM soft); production requires deploying NIM yourself or NVIDIA AI Enterprise.",
          "docs": "https://forums.developer.nvidia.com/t/request-for-nvidia-nim-api-rate-limit-credits-increase/375762"
        },
        {
          "text": "The real reasoning switch is chat_template_kwargs (extra_body): enable_thinking (Qwen-style), medium_effort (Nemotron-3), reasoning_strength (muse-glimmer). reasoning_effort is accepted by the schema but silently ignored on some models.",
          "docs": "https://docs.api.nvidia.com/nim/reference/nvidia-nemotron-3-ultra-550b-a55b"
        },
        {
          "text": "Hosted temperature range is 0-1 (default 0.95), not 0-2. Old catalog entries (e.g. meta/llama2-70b) still listed but 404.",
          "docs": "https://docs.api.nvidia.com/nim/reference/meta-muse-glimmer-30b"
        }
      ],
      "offerings": [
        {
          "model": "deepseek/deepseek-v4-pro",
          "wire_id": "deepseek-ai/deepseek-v4-pro",
          "endpoint": "v1-chat-completions",
          "status": "retired",
          "status_date": "2026-08-19",
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://docs.api.nvidia.com/nim/reference/deepseek-ai-deepseek-v4-pro",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Modes non-think/think-high/think-max via chat_template_kwargs; reasoning_effort reportedly silently ignored on this model (forum) — verify. Removed from the live NIM catalog 2026-08-19 (only deepseek-v4-flash-0731 remains); no replacement id.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.api.nvidia.com/nim/reference/deepseek-ai-deepseek-v4-pro",
              "verified": "2026-08-19"
            }
          }
        },
        {
          "model": "google/gemma-4",
          "wire_id": "google/gemma-4-31b-it",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": false
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Gemma 4 thinking is configurable upstream (think token); NIM control not documented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://build.nvidia.com/google/gemma-4-31b-it",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "meta/llama-3-3-70b",
          "wire_id": "meta/llama-3.3-70b-instruct",
          "endpoint": "v1-chat-completions",
          "status": "retired",
          "status_date": "2026-08-21",
          "limits": {
            "context": 131072,
            "output": null,
            "source": {
              "url": "https://docs.api.nvidia.com/nim/reference/meta-llama-3-3-70b-instruct",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "none",
            "mandatory": false,
            "default": "off",
            "notes": "Llama 3.3 is not a reasoning model; no reasoning parameter on this surface. Removed from the live NIM catalog 2026-08-21.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://build.nvidia.com/meta/llama-3_3-70b-instruct",
              "verified": "2026-08-19"
            }
          }
        },
        {
          "model": "meta/muse-glimmer-30b",
          "wire_id": "meta/muse-glimmer-30b",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 131072,
            "output": null,
            "source": {
              "url": "https://docs.api.nvidia.com/nim/reference/meta-muse-glimmer-30b",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.api.nvidia.com/nim/reference/meta-muse-glimmer-30b",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "minimal",
                "low",
                "medium",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Model-dependent; chat_template_kwargs.reasoning_strength also documented — verify."
            }
          }
        },
        {
          "model": "minimax/minimax-m3",
          "wire_id": "minimaxai/minimax-m3",
          "endpoint": "v1-chat-completions",
          "status": "retired",
          "status_date": "2026-09-10",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "MiniMax M-series cannot disable thinking upstream; NIM control not documented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://build.nvidia.com/minimaxai/minimax-m3",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium",
              "notes": "Values follow upstream; NIM docs — verify."
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-6",
          "wire_id": "moonshotai/kimi-k2.6",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": false
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Kimi chat-template thinking flag; NIM per-model docs — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://build.nvidia.com/moonshotai/kimi-k2.6",
              "verified": "2026-08-28"
            },
            "toggle": {
              "param": "chat_template_kwargs.thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "moonshot/kimi-k3",
          "wire_id": "moonshotai/kimi-k3",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": false
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Kimi chat-template thinking flag; NIM per-model docs — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://build.nvidia.com/moonshotai/kimi-k3",
              "verified": "2026-08-28"
            },
            "toggle": {
              "param": "chat_template_kwargs.thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "nvidia/nemotron-3-5-lightning",
          "wire_id": "nvidia/nemotron-3.5-lightning-30b-a3b",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://docs.api.nvidia.com/nim/reference/nvidia-nemotron-3-5-lightning-30b-a3b",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Nemotron-3 family convention is chat_template_kwargs (extra_body); not documented for 3.5 Lightning on the reference page — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.api.nvidia.com/nim/reference/nvidia-nemotron-3-5-lightning-30b-a3b",
              "verified": "2026-08-20"
            },
            "toggle": {
              "param": "chat_template_kwargs.enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "nvidia/nemotron-3-super",
          "wire_id": "nvidia/nemotron-3-super-120b-a12b",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": false
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "chat_template_kwargs.medium_effort: true adds a medium effort tier.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://build.nvidia.com/nvidia/nemotron-3-super-120b-a12b",
              "verified": "2026-08-28"
            },
            "toggle": {
              "param": "chat_template_kwargs.enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "nvidia/nemotron-3-ultra",
          "wire_id": "nvidia/nemotron-3-ultra-550b-a55b",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://developer.nvidia.com/blog/nvidia-nemotron-3-ultra-powers-faster-more-efficient-reasoning-for-long-running-agents/",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "chat_template_kwargs.medium_effort: true adds a medium effort tier.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.api.nvidia.com/nim/reference/nvidia-nemotron-3-ultra-550b-a55b",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "chat_template_kwargs.enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "openai/gpt-oss-120b",
          "wire_id": "openai/gpt-oss-120b",
          "endpoint": "v1-chat-completions",
          "status": "retired",
          "status_date": "2026-09-04",
          "limits": {
            "context": 131072,
            "output": null,
            "source": {
              "url": "https://docs.api.nvidia.com/nim/reference/openai-gpt-oss-120b",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.api.nvidia.com/nim/reference/openai-gpt-oss-120b",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "chat_template_kwargs.enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "openai/gpt-oss-20b",
          "wire_id": "openai/gpt-oss-20b",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": false
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://build.nvidia.com/openai/gpt-oss-20b",
              "verified": "2026-08-28"
            },
            "toggle": {
              "param": "chat_template_kwargs.enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "poolside/laguna-xs-2-1",
          "wire_id": "poolside/laguna-xs-2.1",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://docs.poolside.ai",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "No reasoning parameters documented upstream — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://build.nvidia.com/poolside/laguna-xs-2-1",
              "verified": "2026-08-20"
            }
          }
        },
        {
          "model": "stepfun/step-3-7-flash",
          "wire_id": "stepfun-ai/step-3.7-flash",
          "endpoint": "v1-chat-completions",
          "status": "retired",
          "status_date": "2026-08-28",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Upstream has three reasoning levels + Advisor Mode; NIM control not documented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://build.nvidia.com/stepfun-ai/step-3.7-flash",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        }
      ]
    },
    {
      "id": "ollama-cloud",
      "name": "Ollama Cloud",
      "kind": "first_party",
      "urls": {
        "docs": "https://docs.ollama.com/cloud",
        "console": "https://ollama.com/settings/keys",
        "pricing": "https://ollama.com/cloud"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "OLLAMA_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create keys at ollama.com/settings/keys (they do not currently expire).",
          "docs": "https://docs.ollama.com/api/authentication"
        }
      ],
      "endpoints": [
        {
          "id": "v1-chat-completions",
          "base_url": "https://ollama.com",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Native API at https://ollama.com/api (chat/generate/tags) uses the think parameter (low/medium/high/max); gpt-oss models accept only low/medium/high and cannot disable thinking. An Anthropic-compatible API is also documented. https://ollama.com/api/v1 does NOT exist (404).",
          "docs": "https://docs.ollama.com/capabilities/thinking"
        },
        {
          "text": "The OpenAI-compatible /v1 accepts reasoning_effort (high/medium/low/max/none) and reasoning.effort. Reasoning returns in message.thinking on the native API — not reasoning_content. Cloud embeddings are unverified: /api/embed may not be authorized for cloud keys.",
          "docs": "https://docs.ollama.com/api/openai-compatibility"
        },
        {
          "text": "Usage levels (Low, Medium, High, Extra High) weight billing — no per-Mtok pricing. Full catalog at ollama.com/search?c=cloud (also gemma4, nemotron-3 family, mistral-large-3 — the only non-thinking cloud model).",
          "docs": "https://ollama.com/cloud"
        },
        {
          "text": "Model IDs dropped the -cloud suffix (verified 2026-08-19); DeepSeek and Qwen entries carry snapshot/variant suffixes (:0731/:0813/:397b/:preview) and full-catalog drift is visible in the daily sync.",
          "docs": "https://ollama.com/search?c=cloud"
        }
      ],
      "plan": {
        "price_usd": 20,
        "period": "monthly",
        "quota": "Free: 1 concurrent model; Pro $20: 3; Max $100: 10. Sessions reset every 5 hours plus weekly limits; usage weighted by model cost level; extra balance purchasable on Pro/Max.",
        "notes": "Zero data retention; no training on prompts; US-primary hosting.",
        "docs": "https://ollama.com/cloud"
      },
      "offerings": [
        {
          "model": "deepseek/deepseek-v4-flash",
          "wire_id": "deepseek-v4-flash:0731",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-pro",
          "wire_id": "deepseek-v4-pro:0813",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "google/gemma-4",
          "wire_id": "gemma4:31b",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://ollama.com/library/gemma4",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Gemma 4 has configurable thinking modes via a think token in the chat template; native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://ollama.com/library/gemma4",
              "verified": "2026-08-20"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "minimax/minimax-m2-7",
          "wire_id": "minimax-m2.7",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 204800,
            "output": null,
            "source": {
              "url": "https://platform.minimax.io/docs/guides/text-generation",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "minimax/minimax-m3",
          "wire_id": "minimax-m3",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-21"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "mistral/mistral-large-3",
          "wire_id": "mistral-large-3:675b",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://ollama.com/library/mistral-large-3",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "none",
            "mandatory": false,
            "default": "off",
            "notes": "Mistral Large 3 is the only non-thinking Ollama Cloud model (no thinking capability tag); library page also lists the tag as mistral-large-3:675b-cloud while the live /v1/models id is mistral-large-3:675b.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://ollama.com/library/mistral-large-3",
              "verified": "2026-08-20"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-6",
          "wire_id": "kimi-k2.6",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-7-code",
          "wire_id": "kimi-k2.7-code",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Native API returns message.thinking; /v1 return shape varies — verify. Interleaved thinking; preserve_thinking supported.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "moonshot/kimi-k3",
          "wire_id": "kimi-k3",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Requires Pro/Max and consumes extra credits; reference per-Mtok pricing $3/$15 (in/output). Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "nvidia/nemotron-3-super",
          "wire_id": "nemotron-3-super",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-21"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "nvidia/nemotron-3-ultra",
          "wire_id": "nemotron-3-ultra",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "openai/gpt-oss-120b",
          "wire_id": "gpt-oss:120b",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 131072,
            "output": null,
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "openai/gpt-oss-20b",
          "wire_id": "gpt-oss:20b",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 131072,
            "output": null,
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "zai/glm-5-1",
          "wire_id": "glm-5.1",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-21"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "zai/glm-5-2",
          "wire_id": "glm-5.2",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "max"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "zai/glm-5-3",
          "wire_id": "glm-5.3",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "GLM-5.3 thinks always (thinking.type disabled errors upstream); effort sets depth. Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        },
        {
          "model": "zai/glm-5-3-flash",
          "wire_id": "glm-5.3-flash",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "GLM-5.3 Flash cannot disable thinking upstream; effort sets depth. Native API returns message.thinking; /v1 return shape varies — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.ollama.com/cloud",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high",
              "notes": "Default level not documented; verify."
            }
          }
        }
      ]
    },
    {
      "id": "openai",
      "name": "OpenAI",
      "kind": "first_party",
      "urls": {
        "docs": "https://developers.openai.com/api/reference/overview/",
        "console": "https://platform.openai.com",
        "status": "https://status.openai.com"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "OPENAI_API_KEY"
          ],
          "key_prefix": "sk-",
          "extra_headers": {},
          "getting_credentials": "platform.openai.com → API Keys. Project keys (sk-proj-) scope to a project; send OpenAI-Organization / OpenAI-Project when a key spans orgs/projects.",
          "docs": "https://developers.openai.com/api/reference/overview/"
        },
        {
          "id": "chatgpt-oauth",
          "type": "oauth",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [],
          "extra_headers": {},
          "getting_credentials": "codex login (or --device-auth). Tokens persist in ~/.codex/auth.json (or OS keyring); refresh tokens rotate and are effectively single-use. Billed against ChatGPT Plus/Pro/Business plans (5-hour + weekly windows), not API credits.",
          "docs": "https://learn.chatgpt.com/docs/auth",
          "flow": "authorization_code_pkce or device code at auth.openai.com (public client; first-party clients only)",
          "scopes": [
            "openid",
            "profile",
            "email",
            "offline_access"
          ]
        }
      ],
      "endpoints": [
        {
          "id": "v1-chat-completions",
          "base_url": "https://api.openai.com",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "v1-responses",
          "base_url": "https://api.openai.com",
          "path": "/v1/responses",
          "protocol": "openai-responses"
        },
        {
          "id": "codex-backend",
          "base_url": "https://chatgpt.com",
          "path": "/backend-api/codex/responses",
          "protocol": "openai-responses",
          "auth": "chatgpt-oauth"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "embeddings",
        "files",
        "batch",
        "fine_tuning",
        "realtime",
        "image_gen",
        "audio"
      ],
      "quirks": [
        {
          "text": "Assistants API sunsets 2026-08-26; migrate to Responses. Reasoning models require max_completion_tokens (chat) instead of max_tokens.",
          "docs": "https://developers.openai.com/api/docs/guides/migrate-to-responses"
        },
        {
          "text": "ChatGPT-plan requests go to chatgpt.com/backend-api/codex (not api.openai.com) with Bearer tokens plus ChatGPT-Account-ID and originator headers; OpenAI allow-lists first-party originators server-side — third-party clients mimicking Codex work but are officially unsupported (ban reports exist). Enterprise access tokens (CODEX_ACCESS_TOKEN) are the supported programmatic path.",
          "docs": "https://learn.chatgpt.com/docs/auth"
        },
        {
          "text": "Subscription quotas: rolling 5-hour windows shared by local messages and cloud tasks plus weekly caps (Plus/Business: gpt-5.6-luna 50-280 local messages/5h; Pro 5x-20x that; flexible-pricing enterprises scale with credits — credits per 1M tokens: Sol 125/750 in/out, Terra 50/300, Luna 5/30). GPT-5.4/5.4-mini retire from ChatGPT-plan access 2026-08-31.",
          "docs": "https://chatgpt.com/codex/pricing/"
        }
      ],
      "offerings": [
        {
          "model": "openai/gpt-5",
          "wire_id": "gpt-5",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 10,
            "cache_read": 0.13,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://developers.openai.com/api/docs/guides/latest-model",
              "verified": "2026-08-18"
            }
          },
          "limits": {
            "context": 400000,
            "output": 128000,
            "source": {
              "url": "https://developers.openai.com/api/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "No 'none': reasoning cannot be fully disabled. minimal-low-medium-high; default medium.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://developers.openai.com/api/docs/guides/reasoning",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5",
          "wire_id": "gpt-5",
          "endpoint": "v1-responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 10,
            "cache_read": 0.13,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://developers.openai.com/api/docs/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 400000,
            "output": 128000,
            "source": {
              "url": "https://developers.openai.com/api/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "hidden",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://developers.openai.com/api/docs/guides/reasoning",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-1",
          "wire_id": "gpt-5.1",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 10,
            "cache_read": 0.125,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://developers.openai.com/api/docs/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 400000,
            "output": 128000,
            "source": {
              "url": "https://developers.openai.com/api/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Defaults to none; tool calls supported at all efforts.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://developers.openai.com/api/docs/guides/reasoning",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "none"
            }
          }
        },
        {
          "model": "openai/gpt-5-1",
          "wire_id": "gpt-5.1",
          "endpoint": "v1-responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 10,
            "cache_read": 0.125,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://developers.openai.com/api/docs/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 400000,
            "output": 128000,
            "source": {
              "url": "https://developers.openai.com/api/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://developers.openai.com/api/docs/guides/reasoning",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "none"
            }
          }
        },
        {
          "model": "openai/gpt-5-5",
          "wire_id": "gpt-5.5",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 30,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "notes": "Prompts >272K billed 2x input / 1.5x output; GPT-5.5-Pro variant is 30/180.",
            "source": {
              "url": "https://developers.openai.com/api/docs/models/gpt-5.5",
              "verified": "2026-08-21"
            }
          },
          "limits": {
            "context": 1050000,
            "output": 128000,
            "source": {
              "url": "https://developers.openai.com/api/docs/models/gpt-5.5",
              "verified": "2026-08-21"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://developers.openai.com/api/docs/models/gpt-5.5",
              "verified": "2026-08-21"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-5",
          "wire_id": "gpt-5.5",
          "endpoint": "v1-responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 30,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "notes": "Prompts >272K billed 2x input / 1.5x output; GPT-5.5-Pro variant is 30/180.",
            "source": {
              "url": "https://developers.openai.com/api/docs/models/gpt-5.5",
              "verified": "2026-08-21"
            }
          },
          "limits": {
            "context": 1050000,
            "output": 128000,
            "source": {
              "url": "https://developers.openai.com/api/docs/models/gpt-5.5",
              "verified": "2026-08-21"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://developers.openai.com/api/docs/models/gpt-5.5",
              "verified": "2026-08-21"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "gpt-5.6",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 12,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "notes": "Plain gpt-5.6 id bills as Terra; variant prices: sol 5/30, luna 0.2/1.2, tiered over 200k — verify mapping.",
            "source": {
              "url": "https://developers.openai.com/api/docs/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 400000,
            "output": 128000,
            "source": {
              "url": "https://developers.openai.com/api/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "tools + reasoning_effort conflict on Chat Completions: use effort none or the Responses endpoint.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://learn.microsoft.com/en-us/azure/foundry/openai/how-to/reasoning",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "gpt-5.6-luna",
          "endpoint": "codex-backend",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.2,
            "output": 1.2,
            "cache_read": 0.02,
            "cache_write": null,
            "free": false,
            "notes": "Short-ctx tier; long-ctx doubles.",
            "source": {
              "url": "https://developers.openai.com/api/docs/pricing",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Fast Luna variant; highest Plus/Business 5h message counts (50-280).",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [],
            "source": {
              "url": "https://learn.chatgpt.com/docs/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "gpt-5.6",
          "endpoint": "v1-responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 12,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "notes": "Plain gpt-5.6 id bills as Terra; variant prices: sol 5/30, luna 0.2/1.2, tiered over 200k — verify mapping.",
            "source": {
              "url": "https://developers.openai.com/api/docs/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 400000,
            "output": 128000,
            "source": {
              "url": "https://developers.openai.com/api/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "'max' effort only on Responses. reasoning.mode standard|pro and reasoning.context auto|current_turn|all_turns (all_turns default) are 5.6-only.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://developers.openai.com/api/docs/guides/reasoning",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "gpt-5.6-sol",
          "endpoint": "codex-backend",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 30,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "notes": "Short-ctx tier; long-ctx doubles.",
            "source": {
              "url": "https://developers.openai.com/api/docs/pricing",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Highest-effort Sol variant; subscription auth only.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [],
            "source": {
              "url": "https://learn.chatgpt.com/docs/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "gpt-5.6-terra",
          "endpoint": "codex-backend",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 12,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "notes": "Short-ctx tier; long-ctx doubles.",
            "source": {
              "url": "https://developers.openai.com/api/docs/pricing",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Balanced Terra variant.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [],
            "source": {
              "url": "https://learn.chatgpt.com/docs/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/o3",
          "wire_id": "o3",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 8,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://developers.openai.com/api/docs/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 200000,
            "output": 100000,
            "source": {
              "url": "https://developers.openai.com/api/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://developers.openai.com/api/docs/guides/reasoning",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/o3",
          "wire_id": "o3",
          "endpoint": "v1-responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 8,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://developers.openai.com/api/docs/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 200000,
            "output": 100000,
            "source": {
              "url": "https://developers.openai.com/api/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "hidden",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://developers.openai.com/api/docs/guides/reasoning",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        }
      ]
    },
    {
      "id": "opencode-go",
      "name": "OpenCode Go",
      "kind": "subscription",
      "urls": {
        "docs": "https://opencode.ai/docs/go/",
        "console": "https://opencode.ai/go"
      },
      "auth": [
        {
          "id": "plan-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "OPENCODE_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Subscribe at opencode.ai/go; uses the OpenCode Zen API key. Provider IDs are opencode-go/<model>.",
          "docs": "https://opencode.ai/docs/go/"
        }
      ],
      "endpoints": [
        {
          "id": "go-responses",
          "base_url": "https://opencode.ai",
          "path": "/zen/go/v1/responses",
          "protocol": "openai-responses"
        },
        {
          "id": "go-chat-completions",
          "base_url": "https://opencode.ai",
          "path": "/zen/go/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "go-messages",
          "base_url": "https://opencode.ai",
          "path": "/zen/go/v1/messages",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Per-model-family protocol routing: /responses for Grok/GPT, /chat/completions for GLM/Kimi/DeepSeek, /messages for MiniMax/Qwen. Offerings verified against the live GET /zen/go/v1/models list (2026-08-19).",
          "docs": "https://opencode.ai/docs/go/"
        }
      ],
      "plan": {
        "price_usd": 10,
        "period": "monthly",
        "quota": "$12 per 5 hours, $30/week, $60/month",
        "notes": "$5 first month then $10/mo, on top of the Zen console (same API key mechanism); open coding models only — Grok 4.5, GPT 5.6 Luna, GLM-5.x, Kimi K3/K2.x, MiniMax M3/M2.7, Qwen3.x, DeepSeek V4, MiMo, Hy3; optional Zen-balance fallback. Routes per protocol: /responses for Grok/GPT, /chat/completions for GLM/Kimi/DeepSeek, /messages (Anthropic) for MiniMax/Qwen.",
        "docs": "https://opencode.ai/docs/go/"
      },
      "offerings": [
        {
          "model": "alibaba/qwen3-7-max",
          "wire_id": "qwen3.7-max",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; thinking mode exists upstream, control undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "alibaba/qwen3-7-plus",
          "wire_id": "qwen3.7-plus",
          "endpoint": "go-messages",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; DashScope wire is enable_thinking true/false — verify the exact placement on this Anthropic-protocol endpoint.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "alibaba/qwen3-8-flash",
          "wire_id": "qwen3.8-flash",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "off",
            "notes": "Qwen flash tiers default thinking off; NEAR documents chat_template_kwargs.enable_thinking for Qwen — verify on this gateway.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-28"
            },
            "toggle": {
              "param": "chat_template_kwargs.enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "alibaba/qwen3-8-max",
          "wire_id": "qwen3.8-max",
          "endpoint": "go-messages",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; DashScope wire is enable_thinking true/false — verify the exact placement on this Anthropic-protocol endpoint.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-20"
            },
            "toggle": {
              "param": "enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-flash",
          "wire_id": "deepseek-v4-flash",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from DeepSeek's native offering. Thinking enabled by default, default effort high; native values low/high/max.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "high",
                "max"
              ],
              "default": "high"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-pro",
          "wire_id": "deepseek-v4-pro",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from DeepSeek's native offering. Thinking enabled by default, default effort high; native values low/high/max.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "high",
                "max"
              ],
              "default": "high"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "meituan/longcat-2-0",
          "wire_id": "longcat-2.0",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; LongCat toggles thinking via thinking-budget chat-template commands (min 1024 when on, 0 disables) — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-28"
            },
            "toggle": {
              "param": "thinking_budget",
              "on": "1024",
              "off": "0"
            }
          }
        },
        {
          "model": "meta/muse-spark-1-2",
          "wire_id": "muse-spark-1.2-contributor",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Meta's native offering. Contributor variant.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "meta/muse-spark-1-2",
          "wire_id": "muse-spark-1.2",
          "endpoint": "go-chat-completions",
          "status": "retired",
          "status_date": "2026-08-20",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Meta's native offering. Removed from the Go live list 2026-08-20 (only muse-spark-1.2-contributor remains).",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "meta/muse-spark-1-3",
          "wire_id": "muse-spark-1.3-contributor",
          "endpoint": "go-chat-completions",
          "status": "preview",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Contributor SKU of Muse Spark 1.3; always-on reasoning per Meta's native format — verify on this gateway.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-09-04"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "minimax/minimax-m2-5",
          "wire_id": "minimax-m2.5",
          "endpoint": "go-messages",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from MiniMax's native M2.x offering — thinking cannot be turned off.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            }
          }
        },
        {
          "model": "minimax/minimax-m2-7",
          "wire_id": "minimax-m2.7",
          "endpoint": "go-messages",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 204800,
            "output": null,
            "source": {
              "url": "https://platform.minimax.io/docs/guides/text-generation",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from MiniMax's native M2.x offering — thinking cannot be turned off.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            }
          }
        },
        {
          "model": "minimax/minimax-m3",
          "wire_id": "minimax-m3",
          "endpoint": "go-messages",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "adaptive",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from MiniMax's native offering, mapped to the thinking object. thinking.type is adaptive|disabled (no 'enabled'); adaptive lets the model decide.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "adaptive",
              "off": "disabled"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-5",
          "wire_id": "kimi-k2.5",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Moonshot's native offering. k2.5 has no keep param — round-trip rule not documented.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-21"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-6",
          "wire_id": "kimi-k2.6",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Moonshot's native offering.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [
              "temperature",
              "top_p",
              "n",
              "presence_penalty",
              "frequency_penalty"
            ],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-7-code",
          "wire_id": "kimi-k2.7-code",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Moonshot's native offering — thinking.type accepts ONLY enabled.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [
              "temperature",
              "top_p",
              "n",
              "presence_penalty",
              "frequency_penalty"
            ],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": ""
            }
          }
        },
        {
          "model": "moonshot/kimi-k3",
          "wire_id": "kimi-k3",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; kimi speaks thinking.type on chat surfaces per community evidence — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [
              "temperature",
              "top_p",
              "n",
              "presence_penalty",
              "frequency_penalty"
            ],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "gpt-5.6-luna",
          "endpoint": "go-responses",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from OpenAI's native offering. Fast Luna variant.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "tencent/hy3",
          "wire_id": "hy3-preview",
          "endpoint": "go-chat-completions",
          "status": "preview",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Upstream defaults to no-think with low/high CoT modes.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "tencent/hy3",
          "wire_id": "hy3",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 262144,
            "output": 131072,
            "source": {
              "url": "https://www.tencentcloud.com/document/product/1300/80695",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; Hunyuan Hy3 defaults to no-think with low/high CoT modes — wire params here not documented, verify. Max input 192k.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-20"
            }
          }
        },
        {
          "model": "tencent/hy4",
          "wire_id": "hy4-preview",
          "endpoint": "go-chat-completions",
          "status": "preview",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Hy3 uses a no-think default with low/high CoT modes upstream; Hy4 preview control undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "none",
                "low",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "xai/grok-4-5",
          "wire_id": "grok-4.5",
          "endpoint": "go-responses",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from xAI's native offering. xhigh is silently treated as high on 4.5.",
            "returns": "hidden",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "xai/grok-4-6",
          "wire_id": "grok-4.6",
          "endpoint": "go-responses",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from xAI's native offering.",
            "returns": "reasoning_content",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-21"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "xiaomi/mimo-v2-5",
          "wire_id": "mimo-v2.5",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://mimo.mi.com/docs/en-US/quickstart",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; upstream MiMo returns reasoning_content which must be passed back in tool loops — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-20"
            }
          }
        },
        {
          "model": "xiaomi/mimo-v2-5-pro",
          "wire_id": "mimo-v2.5-pro",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://mimo.mi.com/docs/en-US/quickstart",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; upstream MiMo returns reasoning_content which must be passed back in tool loops — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-20"
            }
          }
        },
        {
          "model": "zai/glm-5-1",
          "wire_id": "glm-5.1",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Z.ai's native offering.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-21"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "zai/glm-5-2",
          "wire_id": "glm-5.2",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; upstream GLM wire on the openai-chat surface is thinking.type — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "zai/glm-5-3",
          "wire_id": "glm-5.3",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Z.ai's native offering — forced thinking, thinking.type enabled and disabled both return 400.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-19"
            }
          }
        },
        {
          "model": "zai/glm-5-3-flash",
          "wire_id": "glm-5.3-flash",
          "endpoint": "go-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; GLM-5.3 Flash thinks always (enabled-only upstream) — verify on this gateway.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/go/v1/models",
              "verified": "2026-08-28"
            }
          }
        }
      ]
    },
    {
      "id": "opencode-zen",
      "name": "OpenCode Zen",
      "kind": "aggregator",
      "urls": {
        "docs": "https://opencode.ai/docs/zen/",
        "console": "https://opencode.ai/auth"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "OPENCODE_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create a key at opencode.ai/auth (in the TUI: /connect). ~90+ curated models billed at cost, pay-as-you-go.",
          "docs": "https://opencode.ai/docs/providers/"
        }
      ],
      "endpoints": [
        {
          "id": "chat-completions",
          "base_url": "https://opencode.ai",
          "path": "/zen/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "responses",
          "base_url": "https://opencode.ai",
          "path": "/zen/v1/responses",
          "protocol": "openai-responses"
        },
        {
          "id": "messages",
          "base_url": "https://opencode.ai",
          "path": "/zen/v1/messages",
          "protocol": "anthropic-messages"
        },
        {
          "id": "gemini-style",
          "base_url": "https://opencode.ai",
          "path": "/zen/v1/models/{model-id}",
          "protocol": "google-generate-content"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Four protocol surfaces; models are assigned per surface (Responses for GPT/Grok, Anthropic Messages for Claude/Qwen, chat-completions for DeepSeek/MiniMax/GLM/Kimi, Gemini-style for Google models). No documented reasoning normalization — controls follow each upstream provider's wire format per surface; community-verified: thinking {type: enabled} and reasoning_effort (low/high/max) pass through for DeepSeek V4 on the chat surface. Offerings verified against the live GET /zen/v1/models list (2026-08-19).",
          "docs": "https://opencode.ai/docs/zen/"
        }
      ],
      "offerings": [
        {
          "model": "allenai/olmo-3-32b-think",
          "wire_id": "olmo-3-32b-think",
          "endpoint": "chat-completions",
          "status": "retired",
          "status_date": "2026-08-27",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; OLMo 3 Think is an always-thinking long-CoT model — round-trip rule not documented; verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-21"
            }
          }
        },
        {
          "model": "anthropic/claude-fable-5",
          "wire_id": "claude-fable-5",
          "endpoint": "messages",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 10,
            "output": 50,
            "cache_read": 1,
            "cache_write": 12.5,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": true,
            "default": "adaptive",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Anthropic's native offering. thinking.type disabled→400 on fable-5 (mandatory=true); effort changes invalidate prompt-cache breakpoints.",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k",
              "prefill"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "output_config.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-fable-5-1",
          "wire_id": "claude-fable-5-1",
          "endpoint": "messages",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 10,
            "output": 50,
            "cache_read": 0.25,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": true,
            "default": "adaptive",
            "notes": "Cloned from Anthropic's native offering. thinking.type disabled errors on the fable line; effort changes invalidate prompt-cache breakpoints.",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k",
              "prefill"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-09-04"
            },
            "effort": {
              "param": "output_config.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-haiku-4-5",
          "wire_id": "claude-haiku-4-5",
          "endpoint": "messages",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1,
            "output": 5,
            "cache_read": 0.1,
            "cache_write": 1.25,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "budget",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Anthropic's native offering.",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "budget": {
              "param": "thinking.budget_tokens",
              "min": 1024,
              "zero_means_off": false,
              "special_values": {},
              "constraint": "must be < max_tokens"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-4-6",
          "wire_id": "claude-opus-4-6",
          "endpoint": "messages",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "adaptive",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Anthropic's native offering. 4.x Opus accepts thinking.type disabled at effort ≤ high like the 5-series — verify.",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k",
              "prefill"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "output_config.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-4-7",
          "wire_id": "claude-opus-4-7",
          "endpoint": "messages",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "adaptive",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Anthropic's native offering. 4.x Opus accepts thinking.type disabled at effort ≤ high like the 5-series — verify.",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k",
              "prefill"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "output_config.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-4-8",
          "wire_id": "claude-opus-4-8",
          "endpoint": "messages",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "adaptive",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Anthropic's native offering. 4.x Opus accepts thinking.type disabled at effort ≤ high like the 5-series — verify.",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k",
              "prefill"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "output_config.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-5",
          "wire_id": "claude-opus-5",
          "endpoint": "messages",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": 6.25,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "adaptive",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Anthropic's native offering. opus-5/sonnet-5 accept thinking.type disabled only at effort ≤ high; effort changes invalidate prompt-cache breakpoints.",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k",
              "prefill"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "output_config.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-sonnet-5",
          "wire_id": "claude-sonnet-5",
          "endpoint": "messages",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 10,
            "cache_read": 0.2,
            "cache_write": 2.5,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "adaptive",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Anthropic's native offering. opus-5/sonnet-5 accept thinking.type disabled only at effort ≤ high; effort changes invalidate prompt-cache breakpoints.",
            "returns": "thinking_blocks",
            "must_round_trip": "signature",
            "incompatible_with": [
              "temperature",
              "top_p",
              "top_k",
              "prefill"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "output_config.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-flash",
          "wire_id": "deepseek-v4-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.44,
            "output": 1.32,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "notes": "Peak hours; off-peak is half.",
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from DeepSeek's native offering. Thinking enabled by default, default effort high; native values low/high/max.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "high",
                "max"
              ],
              "default": "high"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-pro",
          "wire_id": "deepseek-v4-pro",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.32,
            "output": 3.96,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "notes": "Peak hours; off-peak is half.",
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from DeepSeek's native offering. Thinking enabled by default, default effort high; native values low/high/max.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "high",
                "max"
              ],
              "default": "high"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "google/gemini-3-1-pro",
          "wire_id": "gemini-3.1-pro",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 12,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-21"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Google's native offering — thinkingLevel pass-through on the OpenAI-chat surface is not documented; verify.",
            "returns": "thought_parts",
            "must_round_trip": "thought_signature",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-21"
            },
            "effort": {
              "param": "thinkingLevel",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "google/gemini-3-5-flash",
          "wire_id": "gemini-3.5-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.5,
            "output": 9,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Gemini 3.5 uses thinking_level upstream (minimal/low/medium/high, default medium); thinking_budget deprecated — sending both returns 400. Zen wire behavior — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-5-flash-lite",
          "wire_id": "gemini-3.5-flash-lite",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 2.5,
            "cache_read": 0.03,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Cloned from OpenRouter's unified-effort handling of this model.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-6-flash",
          "wire_id": "gemini-3.6-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.5,
            "output": 7.5,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Cloned from OpenRouter's unified-effort handling of this model.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-7-flash",
          "wire_id": "gemini-3.7-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.5,
            "output": 7.5,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Cloned from OpenRouter's unified-effort handling of this model.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-8-flash",
          "wire_id": "gemini-3.8-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.5,
            "output": 7.5,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Fourth Flash generation; thinking_level upstream.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-09-04"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-flash",
          "wire_id": "gemini-3-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.5,
            "output": 3,
            "cache_read": 0.05,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-21"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Google's native offering — thinkingLevel pass-through on the OpenAI-chat surface is not documented; verify.",
            "returns": "thought_parts",
            "must_round_trip": "thought_signature",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-21"
            },
            "effort": {
              "param": "thinkingLevel",
              "values": [
                "minimal",
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "meituan/longcat-2-0",
          "wire_id": "longcat-2.0",
          "endpoint": "chat-completions",
          "status": "retired",
          "status_date": "2026-08-27",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; LongCat toggles thinking via chat-template commands with thinking_budget — min 1024 when on, 0 disables; exact wire values not documented; verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-21"
            },
            "toggle": {
              "param": "thinking_budget",
              "on": "1024",
              "off": "0"
            }
          }
        },
        {
          "model": "meta/muse-spark-1-2",
          "wire_id": "muse-spark-1.2",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 4.25,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "notes": "Paid tier; Contributor-Free excluded.",
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Meta's native offering.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "meta/muse-spark-1-3",
          "wire_id": "muse-spark-1.3-contributor-free",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0,
            "output": 0,
            "cache_read": null,
            "cache_write": null,
            "free": true,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Free tier of the Contributor SKU; cloned from Meta's native offering (always-on reasoning).",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-09-04"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "meta/muse-spark-1-3",
          "wire_id": "muse-spark-1.3",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 4.25,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-09-10"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Muse Spark 1.3 reasons always (thinking cannot be disabled upstream); effort sets depth. Cloned from Meta's native offering — verify on this gateway.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-09-10"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "minimax/minimax-m2-5",
          "wire_id": "minimax-m2.5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 1.2,
            "cache_read": 0.06,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from MiniMax's native M2.x offering — thinking cannot be turned off.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            }
          }
        },
        {
          "model": "minimax/minimax-m2-7",
          "wire_id": "minimax-m2.7",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 1.2,
            "cache_read": 0.06,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 204800,
            "output": null,
            "source": {
              "url": "https://platform.minimax.io/docs/guides/text-generation",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from MiniMax's native M2.x offering — thinking cannot be turned off.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            }
          }
        },
        {
          "model": "minimax/minimax-m3",
          "wire_id": "minimax-m3",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 1.2,
            "cache_read": 0.06,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "adaptive",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from MiniMax's native offering. thinking.type is adaptive|disabled (no 'enabled'); adaptive lets the model decide.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "adaptive",
              "off": "disabled"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-5",
          "wire_id": "kimi-k2.5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.6,
            "output": 3,
            "cache_read": 0.1,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Moonshot's native offering.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-28"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-6",
          "wire_id": "kimi-k2.6",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.95,
            "output": 4,
            "cache_read": 0.16,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Moonshot's native offering.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [
              "temperature",
              "top_p",
              "n",
              "presence_penalty",
              "frequency_penalty"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-7-code",
          "wire_id": "kimi-k2.7-code",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.95,
            "output": 4,
            "cache_read": 0.19,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from Moonshot's native offering — thinking.type accepts ONLY enabled.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [
              "temperature",
              "top_p",
              "n",
              "presence_penalty",
              "frequency_penalty"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": ""
            }
          }
        },
        {
          "model": "moonshot/kimi-k3",
          "wire_id": "kimi-k3",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 3,
            "output": 15,
            "cache_read": 0.3,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; kimi speaks thinking.type on chat surfaces per community evidence — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [
              "temperature",
              "top_p",
              "n",
              "presence_penalty",
              "frequency_penalty"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "nvidia/nemotron-3-5-lightning",
          "wire_id": "nemotron-3.5-lightning-free",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; upstream NIM wire is chat_template_kwargs.enable_thinking — verify. Free tier on Zen.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-20"
            },
            "toggle": {
              "param": "chat_template_kwargs.enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "nvidia/nemotron-3-ultra",
          "wire_id": "nemotron-3-ultra-free",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; upstream NIM wire is chat_template_kwargs.enable_thinking — verify. Free tier on Zen.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "chat_template_kwargs.enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "openai/gpt-5",
          "wire_id": "gpt-5",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.07,
            "output": 8.5,
            "cache_read": 0.107,
            "cache_write": null,
            "free": false,
            "notes": "Below OpenAI direct — Zen at-cost pass-through.",
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from OpenAI's native offering.",
            "returns": "hidden",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-1",
          "wire_id": "gpt-5.1",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.07,
            "output": 8.5,
            "cache_read": 0.107,
            "cache_write": null,
            "free": false,
            "notes": "Below OpenAI direct — Zen at-cost pass-through.",
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from OpenAI's native offering.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high"
              ],
              "default": "none"
            }
          }
        },
        {
          "model": "openai/gpt-5-4",
          "wire_id": "gpt-5.4-pro",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 30,
            "output": 180,
            "cache_read": 30,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Pro serving tier of GPT-5.4.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "medium",
                "high",
                "xhigh"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-4",
          "wire_id": "gpt-5.4",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2.5,
            "output": 15,
            "cache_read": 0.25,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Cloned from OpenAI's native offering; default effort none.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "none"
            }
          }
        },
        {
          "model": "openai/gpt-5-4-mini",
          "wire_id": "gpt-5.4-mini",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.75,
            "output": 4.5,
            "cache_read": 0.075,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Cloned from OpenAI's native offering; default effort none.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "none"
            }
          }
        },
        {
          "model": "openai/gpt-5-4-nano",
          "wire_id": "gpt-5.4-nano",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.2,
            "output": 1.25,
            "cache_read": 0.02,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Cloned from OpenAI's native offering; default effort none.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "none"
            }
          }
        },
        {
          "model": "openai/gpt-5-5",
          "wire_id": "gpt-5.5-pro",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 30,
            "output": 180,
            "cache_read": 30,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Pro serving tier of GPT-5.5; default effort high.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "medium",
                "high",
                "xhigh"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "openai/gpt-5-5",
          "wire_id": "gpt-5.5",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 30,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-21"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from OpenAI's native offering.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-21"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "gpt-5.6-luna",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.2,
            "output": 1.2,
            "cache_read": 0.02,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from OpenAI's native offering. Fast Luna variant.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "gpt-5.6-sol",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 30,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "notes": "Doubles over the 272k tier.",
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from OpenAI's native offering. Highest-effort Sol variant.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "gpt-5.6-terra",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 12,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from OpenAI's native offering. Balanced Terra variant.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-6",
          "wire_id": "gpt-6-astra",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 10,
            "output": 50,
            "cache_read": 1,
            "cache_write": 12.5,
            "free": false,
            "notes": "Short-context tier; over 272K tokens $20/$75.",
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-09-10"
            }
          },
          "limits": {
            "context": 1050000,
            "output": null,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-09-10"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Effort low-max, no none (thinking cannot be disabled); default undocumented — verify.",
            "returns": "reasoning_summary",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [
              "temperature",
              "top_p"
            ],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-09-10"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "xai/grok-4-5",
          "wire_id": "grok-4.5",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 6,
            "cache_read": 0.3,
            "cache_write": null,
            "free": false,
            "notes": "Doubles for prompts over 200k.",
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from xAI's native offering. xhigh is silently treated as high on 4.5.",
            "returns": "hidden",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "xai/grok-4-6",
          "wire_id": "grok-4.6",
          "endpoint": "responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 6,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "notes": "Doubles for prompts over 200k.",
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; cloned from xAI's native offering.",
            "returns": "reasoning_content",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "zai/glm-5-1",
          "wire_id": "glm-5.1",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.4,
            "output": 4.4,
            "cache_read": 0.26,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-21"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; upstream GLM wire on the openai-chat surface is thinking.type — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-21"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "zai/glm-5-2",
          "wire_id": "glm-5.2",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.4,
            "output": 4.4,
            "cache_read": 0.26,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning follows the upstream wire format on this surface; upstream GLM wire on the openai-chat surface is thinking.type — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-08-19"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "zai/glm-5-3",
          "wire_id": "glm-5.3",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.4,
            "output": 4.4,
            "cache_read": 0.26,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-09-10"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "GLM-5.3 thinks always (thinking.type disabled errors upstream); cloned from Z.ai's native offering — verify on this gateway.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-09-10"
            }
          }
        },
        {
          "model": "zai/glm-5-3-flash",
          "wire_id": "glm-5.3-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.15,
            "output": 0.5,
            "cache_read": 0.03,
            "cache_write": null,
            "free": false,
            "notes": "Matches Z.ai's launch-promo pricing.",
            "source": {
              "url": "https://opencode.ai/docs/zen",
              "verified": "2026-09-10"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "GLM-5.3 Flash thinks always (enabled-only upstream); cloned from Z.ai's native offering — verify on this gateway.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://opencode.ai/zen/v1/models",
              "verified": "2026-09-10"
            }
          }
        }
      ]
    },
    {
      "id": "openrouter",
      "name": "OpenRouter",
      "kind": "aggregator",
      "urls": {
        "docs": "https://openrouter.ai/docs/api-reference/overview",
        "console": "https://openrouter.ai/credits",
        "pricing": "https://openrouter.ai/models"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "OPENROUTER_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "openrouter.ai → Keys. Prepaid credits (5.5% purchase fee); optional HTTP-Referer and X-Title attribution headers.",
          "docs": "https://openrouter.ai/docs/api-reference/overview"
        }
      ],
      "endpoints": [
        {
          "id": "chat-completions",
          "base_url": "https://openrouter.ai",
          "path": "/api/v1/chat/completions",
          "protocol": "openai-chat"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Unified reasoning object: {effort | max_tokens, exclude, enabled}. Effort converts to a token budget where needed (Anthropic clamped min 1024 / max 128000). reasoning.exclude strips reasoning from responses. Routing: provider.order/allow_fallbacks/sort/require_parameters/max_price; :nitro and :floor model suffixes.",
          "docs": "https://openrouter.ai/docs/use-cases/reasoning-tokens"
        }
      ],
      "offerings": [
        {
          "model": "alibaba/qwen3-7-flash",
          "wire_id": "qwen/qwen3.7-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.03,
            "output": 0.13,
            "cache_read": 0.006,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "A Thinking Mode exists upstream but the control parameter is undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "alibaba/qwen3-7-plus",
          "wire_id": "qwen/qwen3.7-plus",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.32,
            "output": 1.28,
            "cache_read": 0.064,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "alibaba/qwen3-8-2-4t-a95b",
          "wire_id": "qwen/qwen3.8-2.4t-a95b",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 6,
            "cache_read": 0.25,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Thinking is mandatory (enable_thinking=false errors); open checkpoint documents reasoning_effort xhigh/medium/low with xhigh default.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "xhigh"
            }
          }
        },
        {
          "model": "alibaba/qwen3-8-27b",
          "wire_id": "qwen/qwen3.8-27b",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.45,
            "output": 3.2,
            "cache_read": 0.05,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "alibaba/qwen3-8-flash",
          "wire_id": "qwen/qwen3.8-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.15,
            "output": 0.47,
            "cache_read": 0.016,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Qwen flash defaults thinking off upstream; OR default not documented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "alibaba/qwen3-8-max",
          "wire_id": "qwen/qwen3.8-max-0902",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 6,
            "cache_read": 0.25,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "xhigh"
            }
          }
        },
        {
          "model": "allenai/olmo-3-32b-think",
          "wire_id": "allenai/olmo-3-32b-think",
          "endpoint": "chat-completions",
          "status": "retired",
          "status_date": "2026-09-04",
          "cost": {
            "input": 0.15,
            "output": 0.5,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 65536,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "OLMo 3 Think always thinks; effort none is not offered on this listing.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "amazon/nova-2-lite",
          "wire_id": "amazon/nova-2-lite-v1",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 2.5,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Nova 2 Lite reasons only when enabled: Bedrock reasoningConfig.type defaults to disabled; OR normalizes to its reasoning parameter. maxReasoningEffort low/medium/high upstream.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "low"
            }
          }
        },
        {
          "model": "anthropic/claude-fable-5",
          "wire_id": "anthropic/claude-fable-5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 10,
            "output": 50,
            "cache_read": 1,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "thinking cannot be disabled on fable-5; effort none maps to thinking.type disabled which errors upstream.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-fable-5-1",
          "wire_id": "anthropic/claude-fable-5.1",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 10,
            "output": 50,
            "cache_read": 0.25,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Fable line cannot disable thinking (effort none errored on 5.0); cache hits cut to $0.25 on 5.1.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-haiku-4-5",
          "wire_id": "anthropic/claude-haiku-4.5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1,
            "output": 5,
            "cache_read": 0.1,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-4-6",
          "wire_id": "anthropic/claude-opus-4.6",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-4-7",
          "wire_id": "anthropic/claude-opus-4.7",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-4-8",
          "wire_id": "anthropic/claude-opus-4.8-fast",
          "endpoint": "chat-completions",
          "status": "retired",
          "status_date": "2026-09-04",
          "cost": {
            "input": 10,
            "output": 50,
            "cache_read": 1,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Fast serving mode of Opus 4.8.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-4-8",
          "wire_id": "anthropic/claude-opus-4.8",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-opus-5",
          "wire_id": "anthropic/claude-opus-5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 25,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "anthropic/claude-sonnet-5",
          "wire_id": "anthropic/claude-sonnet-5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 10,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "baidu/ernie-4-5-vl",
          "wire_id": "baidu/ernie-4.5-vl-424b-a47b",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.42,
            "output": 1.25,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 123000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Open weights use the enable_thinking chat-template flag (default on); OR normalizes. Served by NovitaAI.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "bytedance/seed-2-0-code",
          "wire_id": "bytedance-seed/seed-2.0-code",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.5,
            "output": 3,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Ark thinking object enabled/disabled/auto plus reasoning_effort upstream; OR normalizes. Tiered 2x above 128K prompt tokens.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "bytedance/seed-2-0-lite",
          "wire_id": "bytedance-seed/seed-2.0-lite",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.25,
            "output": 2,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Tiered 2x above 128K prompt tokens.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "bytedance/seed-2-0-mini",
          "wire_id": "bytedance-seed/seed-2.0-mini",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.1,
            "output": 0.4,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Upstream documents minimal/low/medium/high effort modes; OR normalizes.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "bytedance/seed-2-1-turbo",
          "wire_id": "bytedance-seed/seed-2-1-turbo",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.5,
            "output": 2.5,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Ark thinking object enabled/disabled/auto upstream; default undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-flash",
          "wire_id": "deepseek/deepseek-v4-flash-0731",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.05,
            "output": 0.1,
            "cache_read": 0.01,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1310720,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Dated 0731 snapshot of deepseek-v4-flash.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-flash",
          "wire_id": "deepseek/deepseek-v4-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.0784,
            "output": 0.1568,
            "cache_read": 0.01568,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-pro",
          "wire_id": "deepseek/deepseek-v4-pro-0813",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.66,
            "output": 1.98,
            "cache_read": 0.022,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Dated 0813 snapshot of deepseek-v4-pro.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "deepseek/deepseek-v4-pro",
          "wire_id": "deepseek/deepseek-v4-pro",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.6,
            "output": 3.2,
            "cache_read": 0.135,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-1-pro",
          "wire_id": "google/gemini-3.1-pro-preview",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 12,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Preview-tagged listing of gemini-3.1-pro.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-5-flash",
          "wire_id": "google/gemini-3.5-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.5,
            "output": 9,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Uses thinking_level upstream (thinking_budget deprecated; sending both returns 400). OR also accepts reasoning_effort.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-5-flash-lite",
          "wire_id": "google/gemini-3.5-flash-lite",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 2.5,
            "cache_read": 0.03,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-6-flash",
          "wire_id": "google/gemini-3.6-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.75,
            "output": 3.75,
            "cache_read": 0.075,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-7-flash",
          "wire_id": "google/gemini-3.7-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.375,
            "output": 1.875,
            "cache_read": 0.0375,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "minimal thinking level errors upstream; thinking cannot be disabled (mandatory).",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-8-flash",
          "wire_id": "google/gemini-3.8-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.75,
            "output": 3.75,
            "cache_read": 0.075,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Fourth Flash generation; thinking_level upstream like the 3.5+ line.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemini-3-flash",
          "wire_id": "google/gemini-3-flash-preview",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.5,
            "output": 3,
            "cache_read": 0.05,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Preview-tagged listing of gemini-3-flash.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemma-4",
          "wire_id": "google/gemma-4-26b-a4b-it",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.07,
            "output": 0.34,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "OR serves the 26B-A4B MoE size; Gemma 4 thinking is configurable upstream (think token / enable_thinking, default on) — verify this surface.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "google/gemma-4",
          "wire_id": "google/gemma-4-31b-it",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.09,
            "output": 0.34,
            "cache_read": 0.05,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "OR serves the 31B dense variant; Gemma 4 thinking is configurable upstream (think token) — verify this surface.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "ibm/granite-4-1-8b",
          "wire_id": "ibm-granite/granite-4.1-8b",
          "endpoint": "chat-completions",
          "status": "retired",
          "status_date": "2026-09-10",
          "cost": {
            "input": 0.05,
            "output": 0.1,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 131072,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "none",
            "mandatory": false,
            "default": "off",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "ibm/granite-4-2-8b",
          "wire_id": "ibm-granite/granite-4.2-8b",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.06,
            "output": 0.25,
            "cache_read": 0.015,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            }
          },
          "limits": {
            "context": 131072,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Granite 4.2 adds reasoning controls over 4.1.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "inception/mercury-2",
          "wire_id": "inception/mercury-2",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.25,
            "output": 0.75,
            "cache_read": 0.025,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 128000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Diffusion LLM: parallel generate-and-refine, 1000+ tok/s. Also accepts 'instant' (near-zero reasoning latency) — a value outside our effort vocabulary. Default undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "inclusionai/ling-3-0-flash",
          "wire_id": "inclusionai/ling-3.0-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.021,
            "output": 0.063,
            "cache_read": 0.0042,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Ant Ling 3.0 thinking control upstream undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "kwaipilot/kat-coder-pro-v2-5",
          "wire_id": "kwaipilot/kat-coder-pro-v2.5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.74,
            "output": 2.96,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "none",
            "mandatory": false,
            "default": "off",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "liquid/lfm-2-5-2-6b",
          "wire_id": "liquid/lfm-2.5-2.6b:free",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0,
            "output": 0,
            "cache_read": null,
            "cache_write": null,
            "free": true,
            "notes": "OR-served free tier of the open 2.6B edge model.",
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 65536,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "LFM2.5 always reasons (cannot be disabled).",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "meituan/longcat-2-0",
          "wire_id": "meituan/longcat-2.0",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 1.2,
            "cache_read": 0.006,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1048756,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Upstream toggles thinking via thinking_budget; OR normalizes through its reasoning parameter — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "meta/llama-3-3-70b",
          "wire_id": "meta-llama/llama-3.3-70b-instruct",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.71,
            "output": 0.71,
            "cache_read": 0.71,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 131072,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "none",
            "mandatory": false,
            "default": "off",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "meta/llama-4-maverick",
          "wire_id": "meta-llama/llama-4-maverick",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.2,
            "output": 0.8,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "meta/muse-glimmer-30b",
          "wire_id": "meta/muse-glimmer-30b",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.35,
            "output": 1.5,
            "cache_read": 0.04,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "meta/muse-spark-1-2",
          "wire_id": "meta/muse-spark-1.2",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 4.25,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "meta/muse-spark-1-3",
          "wire_id": "meta/muse-spark-1.3-contributor",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.1,
            "output": 0.2,
            "cache_read": 0.002,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Contributor-program SKU of Muse Spark 1.3.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "meta/muse-spark-1-3",
          "wire_id": "meta/muse-spark-1.3",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 4.25,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Muse Spark 1.2 lineage: always-on reasoning (effort none returned 400 upstream). 1.3 adds audio+video input; long-context tuned.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "minimax/minimax-m2-5",
          "wire_id": "minimax/minimax-m2.5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.225,
            "output": 0.9,
            "cache_read": 0.06,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "minimax/minimax-m2-7",
          "wire_id": "minimax/minimax-m2.7",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 1.2,
            "cache_read": 0.06,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 204800,
            "output": null,
            "source": {
              "url": "https://platform.minimax.io/docs/guides/text-generation",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "minimax/minimax-m3",
          "wire_id": "minimax/minimax-m3",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 1.2,
            "cache_read": 0.06,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "mistral/mistral-large-3",
          "wire_id": "mistralai/mistral-large-2512",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.5,
            "output": 1.5,
            "cache_read": 0.05,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://docs.mistral.ai/resources/known-limitations",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "none",
            "mandatory": false,
            "default": "off",
            "notes": "No reasoning in OpenRouter supported_parameters for this id; OR id is mistral-large-2512 (Large 3).",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          }
        },
        {
          "model": "mistral/mistral-medium-3-5",
          "wire_id": "mistralai/mistral-medium-3-5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.5,
            "output": 7.5,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-5",
          "wire_id": "moonshotai/kimi-k2.5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.45,
            "output": 2.25,
            "cache_read": 0.07,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-6",
          "wire_id": "moonshotai/kimi-k2.6",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.5415,
            "output": 2.28,
            "cache_read": 0.0912,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "moonshot/kimi-k2-7-code",
          "wire_id": "moonshotai/kimi-k2.7-code",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.71,
            "output": 3.5,
            "cache_read": 0.15,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "moonshot/kimi-k3",
          "wire_id": "moonshotai/kimi-k3",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 3,
            "output": 15,
            "cache_read": 0.3,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "nex-agi/nex-n2-5-mini",
          "wire_id": "nex-agi/nex-n2.5-mini:free",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0,
            "output": 0,
            "cache_read": null,
            "cache_write": null,
            "free": true,
            "notes": "OR-served free tier; no paid listing yet.",
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "N2 lineage reasoning via OR unified parameter — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "nex-agi/nex-n2-5-pro",
          "wire_id": "nex-agi/nex-n2.5-pro:free",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0,
            "output": 0,
            "cache_read": null,
            "cache_write": null,
            "free": true,
            "notes": "OR-served free tier; no paid listing yet.",
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "N2 lineage reasoning via OR unified parameter — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "nex-agi/nex-n2-mini",
          "wire_id": "nex-agi/nex-n2-mini",
          "endpoint": "chat-completions",
          "status": "retired",
          "status_date": "2026-09-10",
          "cost": {
            "input": 0.025,
            "output": 0.1,
            "cache_read": 0.0025,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Qwen3.5-architecture lineage; reasoning via OR unified parameter — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "nex-agi/nex-n2-pro",
          "wire_id": "nex-agi/nex-n2-pro",
          "endpoint": "chat-completions",
          "status": "retired",
          "status_date": "2026-09-10",
          "cost": {
            "input": 0.25,
            "output": 1,
            "cache_read": 0.025,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Qwen3.5-architecture lineage; reasoning via OR unified parameter — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "nvidia/nemotron-3-5-lightning",
          "wire_id": "nvidia/nemotron-3.5-lightning",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.08,
            "output": 0.2,
            "cache_read": 0.04,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "OR id drops the -30b-a3b size suffix.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "nvidia/nemotron-3-nano",
          "wire_id": "nvidia/nemotron-3-nano-30b-a3b",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.05,
            "output": 0.2,
            "cache_read": 0.025,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Upstream toggles via chat_template_kwargs.enable_thinking (default on; off mode decodes greedily); OR normalizes.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "nvidia/nemotron-3-super",
          "wire_id": "nvidia/nemotron-3-super-120b-a12b",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.085,
            "output": 0.4,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "OR id carries the -120b-a12b size suffix.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "nvidia/nemotron-3-ultra",
          "wire_id": "nvidia/nemotron-3-ultra-550b-a55b",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.6,
            "output": 3.6,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "OR id carries the -550b-a55b size suffix.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5",
          "wire_id": "openai/gpt-5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 10,
            "cache_read": 0.125,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-1",
          "wire_id": "openai/gpt-5.1",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.25,
            "output": 10,
            "cache_read": 0.125,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-4",
          "wire_id": "openai/gpt-5.4-pro",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 30,
            "output": 180,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1050000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Deep-reasoning serving tier of GPT-5.4 (Responses API only upstream).",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "medium",
                "high",
                "xhigh"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-4",
          "wire_id": "openai/gpt-5.4",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2.5,
            "output": 15,
            "cache_read": 0.25,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1050000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Default effort is none. Prompts over 272K billed 2x input / 1.5x output for the session.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "none"
            }
          }
        },
        {
          "model": "openai/gpt-5-4-mini",
          "wire_id": "openai/gpt-5.4-mini",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.75,
            "output": 4.5,
            "cache_read": 0.075,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 400000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Default effort is none.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "none"
            }
          }
        },
        {
          "model": "openai/gpt-5-4-nano",
          "wire_id": "openai/gpt-5.4-nano",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.2,
            "output": 1.25,
            "cache_read": 0.02,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 400000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "off",
            "notes": "Default effort is none.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "none"
            }
          }
        },
        {
          "model": "openai/gpt-5-5",
          "wire_id": "openai/gpt-5.5-pro",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 30,
            "output": 180,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1050000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": false,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Deep-reasoning serving tier of GPT-5.5 (Responses/Batch only upstream); OpenAI now points peak-intelligence workloads at gpt-5.6-sol with reasoning.mode=pro.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "medium",
                "high",
                "xhigh"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "openai/gpt-5-5",
          "wire_id": "openai/gpt-5.5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 30,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1050000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "openai/gpt-5.6-luna-pro",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.2,
            "output": 1.2,
            "cache_read": 0.02,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "limits": {
            "context": 1050000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Luna served with the reasoning.mode=pro preset; OR lists the preset as its own SKU.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "openai/gpt-5.6-luna",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.2,
            "output": 1.2,
            "cache_read": 0.02,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "GPT-5.6 cost tier (nano successor). Adds reasoning.mode standard|pro (pro bills at standard rates). Plain openai/gpt-5.6 was retired 2026-08-19; live variants: sol/terra/luna plus -pro and :batch forms.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/docs/use-cases/reasoning-tokens",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "openai/gpt-5.6-sol-pro",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 10,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "limits": {
            "context": 1050000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Sol served with the reasoning.mode=pro preset; OR lists the preset as its own SKU.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "openai/gpt-5.6-sol",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 10,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "limits": {
            "context": 1050000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "GPT-5.6 flagship tier; adds reasoning.mode standard|pro (pro bills at standard rates). Promo pricing at least through 2026-11-21.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "openai/gpt-5.6-terra-pro",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 12,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "limits": {
            "context": 1050000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Terra served with the reasoning.mode=pro preset; OR lists the preset as its own SKU.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-04"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-5-6",
          "wire_id": "openai/gpt-5.6-terra",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 12,
            "cache_read": 0.2,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1050000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "GPT-5.6 mid tier; adds reasoning.mode standard|pro.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-6",
          "wire_id": "openai/gpt-6-astra-pro",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 10,
            "output": 50,
            "cache_read": 1,
            "cache_write": null,
            "free": false,
            "notes": "Short-context tier; over 272K tokens $20/$75.",
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            }
          },
          "limits": {
            "context": 1050000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Astra served with the reasoning.mode=pro preset; OR lists the preset as its own SKU at identical pricing.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-6",
          "wire_id": "openai/gpt-6-astra",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 10,
            "output": 50,
            "cache_read": 1,
            "cache_write": null,
            "free": false,
            "notes": "Short-context tier; over 272K tokens $20/$75.",
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            }
          },
          "limits": {
            "context": 1050000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "OpenAI docs list effort low/medium/high/xhigh/max with no none (thinking cannot be disabled); default undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-09-10"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-oss-120b",
          "wire_id": "openai/gpt-oss-120b",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.03,
            "output": 0.17,
            "cache_read": 0.03,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/gpt-oss-20b",
          "wire_id": "openai/gpt-oss-20b",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.03,
            "output": 0.13,
            "cache_read": 0.03,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "openai/o3",
          "wire_id": "openai/o3",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 8,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "poolside/laguna-s-2-1",
          "wire_id": "poolside/laguna-s-2.1",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.09,
            "output": 0.18,
            "cache_read": 0.009,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "poolside/laguna-xs-2-1",
          "wire_id": "poolside/laguna-xs-2.1",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.06,
            "output": 0.12,
            "cache_read": 0.03,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "sakana/fugu-ultra",
          "wire_id": "sakana/fugu-ultra",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 5,
            "output": 30,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1000000,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Sakana multi-agent-as-a-model; reasoning_effort exposed but semantics undocumented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "stepfun/step-3-5-flash",
          "wire_id": "stepfun/step-3.5-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.1,
            "output": 0.3,
            "cache_read": null,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Upstream reasoning incl. Parallel Thinking.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "minimal",
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "stepfun/step-3-7-flash",
          "wire_id": "stepfun/step-3.7-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.2,
            "output": 1.15,
            "cache_read": 0.04,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "Upstream has three reasoning levels + Advisor Mode.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "tencent/hy3",
          "wire_id": "tencent/hy3",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.132,
            "output": 0.528,
            "cache_read": 0.033,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Upstream defaults to no-think with low/high CoT modes.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "thinkingmachines/inkling",
          "wire_id": "thinkingmachines/inkling",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.95,
            "output": 4.05,
            "cache_read": 0.16,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "thinkingmachines/inkling-small",
          "wire_id": "thinkingmachines/inkling-small",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.45,
            "output": 1.2,
            "cache_read": 0.1,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "upstage/solar-pro-4",
          "wire_id": "upstage/solar-pro4",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.3,
            "output": 1.2,
            "cache_read": 0.006,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 524288,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Upstage's flagship reasoning model; native control parameter undocumented — verify. List pricing shown; 90%-off promo ($0.03/$0.12) through ~2026-09-10.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          }
        },
        {
          "model": "xai/grok-4-5",
          "wire_id": "x-ai/grok-4.5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 6,
            "cache_read": 0.3,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "xai/grok-4-6",
          "wire_id": "x-ai/grok-4.6",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 6,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "xiaomi/mimo-v2-5",
          "wire_id": "xiaomi/mimo-v2.5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.14,
            "output": 0.28,
            "cache_read": 0.0028,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://mimo.mi.com/docs/en-US/quickstart",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "xiaomi/mimo-v2-5-pro",
          "wire_id": "xiaomi/mimo-v2.5-pro",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.435,
            "output": 0.87,
            "cache_read": 0.0036,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://mimo.mi.com/docs/en-US/quickstart",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "zai/glm-4-6",
          "wire_id": "z-ai/glm-4.6",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.5,
            "output": 2,
            "cache_read": 0.1,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "zai/glm-4-7",
          "wire_id": "z-ai/glm-4.7",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.4,
            "output": 1.75,
            "cache_read": 0.08,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "zai/glm-4-7-flash",
          "wire_id": "z-ai/glm-4.7-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.06,
            "output": 0.4,
            "cache_read": 0.01,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 202752,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "GLM-4.7 series still supports thinking.type enabled/disabled (only GLM-5.3+ removed disabled).",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "zai/glm-5-1",
          "wire_id": "z-ai/glm-5.1",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.26,
            "output": 3.96,
            "cache_read": 0.234,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 204800,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": false
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "zai/glm-5-2",
          "wire_id": "z-ai/glm-5.2",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.966,
            "output": 3.036,
            "cache_read": 0.1932,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "none",
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "zai/glm-5-3",
          "wire_id": "z-ai/glm-5.3",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.4,
            "output": 4.4,
            "cache_read": 0.26,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-19"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        },
        {
          "model": "zai/glm-5-3-flash",
          "wire_id": "z-ai/glm-5.3-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.075,
            "output": 0.25,
            "cache_read": 0.015,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "limits": {
            "context": 1310720,
            "output": null,
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "GLM-5.3 Flash cannot disable thinking upstream (enabled-only); OR effort maps to thinking depth.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://openrouter.ai/api/v1/models",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh",
                "max"
              ],
              "default": "medium"
            }
          }
        }
      ]
    },
    {
      "id": "poolside",
      "name": "poolside",
      "kind": "first_party",
      "urls": {
        "docs": "https://docs.poolside.ai",
        "console": "https://platform.poolside.ai"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "POOLSIDE_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create a key at platform.poolside.ai.",
          "docs": "https://docs.poolside.ai"
        }
      ],
      "endpoints": [
        {
          "id": "chat-completions",
          "base_url": "https://inference.poolside.ai",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Free for a limited time; model ids prefixed poolside/.",
          "docs": "https://docs.poolside.ai"
        }
      ],
      "offerings": [
        {
          "model": "poolside/laguna-s-2-1",
          "wire_id": "poolside/laguna-s-2.1",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://docs.poolside.ai",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "No reasoning parameters documented — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.poolside.ai",
              "verified": "2026-08-20"
            }
          }
        },
        {
          "model": "poolside/laguna-xs-2-1",
          "wire_id": "poolside/laguna-xs-2.1",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://docs.poolside.ai",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "No reasoning parameters documented — verify.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.poolside.ai",
              "verified": "2026-08-20"
            }
          }
        }
      ]
    },
    {
      "id": "qwen-coding-plan",
      "name": "Qwen Coding Plan",
      "kind": "subscription",
      "urls": {
        "docs": "https://www.alibabacloud.com/help/en/model-studio/coding-plan",
        "console": "https://modelstudio.console.alibabacloud.com"
      },
      "auth": [
        {
          "id": "plan-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "BAILIAN_CODING_PLAN_API_KEY"
          ],
          "key_prefix": "sk-sp-",
          "extra_headers": {},
          "getting_credentials": "Subscribe in Model Studio; plan keys (sk-sp-...) are separate from normal DASHSCOPE_API_KEY and only work on coding-plan endpoints. Configure tools via OPENAI_BASE_URL=https://coding-intl.dashscope.aliyuncs.com/v1 or the Anthropic-compatible /apps/anthropic route. The older Qwen OAuth free tier was discontinued 2026-04-15.",
          "docs": "https://qwenlm.github.io/qwen-code-docs/en/users/configuration/auth/"
        }
      ],
      "endpoints": [
        {
          "id": "coding-chat",
          "base_url": "https://coding-intl.dashscope.aliyuncs.com",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "coding-anthropic",
          "base_url": "https://coding-intl.dashscope.aliyuncs.com",
          "path": "/apps/anthropic",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "China base: https://coding.dashscope.aliyuncs.com/v1. Third-party models on the allowlist (glm-5, glm-4.7, kimi-k2.5, MiniMax-M2.5) are served through the same endpoints; their reasoning parameters follow the upstream models' conventions — offerings seeded only where wire behavior is verified.",
          "docs": "https://www.alibabacloud.com/help/en/model-studio/coding-plan"
        }
      ],
      "plan": {
        "price_usd": 50,
        "period": "monthly",
        "quota": "Pro $50/mo: 6,000 requests / 5 rolling hours, 45,000/week, 90,000/month; slots restock daily 00:00 UTC+8",
        "notes": "Strict model allowlist: qwen3.7-plus, qwen3.6-plus, qwen3.5-plus, qwen3-coder-next, qwen3-coder-plus, glm-5, glm-4.7, kimi-k2.5, MiniMax-M2.5. Interactive coding-tool use only. Lite discontinued: new subscriptions ended 2026-03-20, renewals ended 2026-04-13. No Max or Team tier exists on the Coding Plan (Token Plan Team is a separate product).",
        "docs": "https://www.alibabacloud.com/help/en/model-studio/coding-plan"
      },
      "offerings": [
        {
          "model": "alibaba/qwen3-7-plus",
          "wire_id": "qwen3.7-plus",
          "endpoint": "coding-chat",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://www.alibabacloud.com/help/en/model-studio/deep-thinking",
              "verified": "2026-08-18"
            },
            "toggle": {
              "param": "enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        }
      ]
    },
    {
      "id": "stepfun",
      "name": "StepFun",
      "kind": "first_party",
      "urls": {
        "docs": "https://platform.stepfun.ai",
        "console": "https://platform.stepfun.com"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "STEPFUN_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create a key on the StepFun platform console (EN platform stepfun.ai, CN stepfun.com).",
          "docs": "https://platform.stepfun.ai"
        }
      ],
      "endpoints": [
        {
          "id": "chat-completions",
          "base_url": "https://api.stepfun.ai",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "EN platform stepfun.ai, CN stepfun.com; step-3 (321B/38B VLM, Jul 2025) is the prior generation.",
          "docs": "https://platform.stepfun.ai"
        },
        {
          "text": "Parallel Thinking (PaCoRe) reasoning on 3.5-flash; 3.7-flash has three reasoning levels + Advisor Mode.",
          "docs": "https://static.stepfun.com/blog/step-3.7-flash/"
        }
      ],
      "offerings": [
        {
          "model": "stepfun/step-3-5-flash",
          "wire_id": "step-3.5-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.1,
            "output": 0.3,
            "cache_read": 0.02,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://platform.stepfun.ai/docs/en/guides/pricing/details",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://static.stepfun.com/blog/step-3.5-flash/",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning incl. Parallel Thinking; wire params not enumerated in EN docs — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://static.stepfun.com/blog/step-3.5-flash/",
              "verified": "2026-08-20"
            }
          }
        },
        {
          "model": "stepfun/step-3-7-flash",
          "wire_id": "step-3.7-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.2,
            "output": 1.15,
            "cache_read": 0.04,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://platform.stepfun.ai/docs/en/guides/pricing/details",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://static.stepfun.com/blog/step-3.7-flash/",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Three reasoning levels (low/medium/high); param presumed reasoning_effort on the OpenAI-compat surface — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://static.stepfun.com/blog/step-3.7-flash/",
              "verified": "2026-08-20"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "medium",
              "notes": "Default level not documented; verify."
            }
          }
        }
      ]
    },
    {
      "id": "synthetic",
      "name": "Synthetic",
      "kind": "subscription",
      "urls": {
        "docs": "https://dev.synthetic.new/",
        "console": "https://synthetic.new",
        "pricing": "https://synthetic.new/pricing"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "SYNTHETIC_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Subscribe at synthetic.new and create an API key in the dashboard; send as Authorization: Bearer SYNTHETIC_API_KEY.",
          "docs": "https://dev.synthetic.new/"
        }
      ],
      "endpoints": [
        {
          "id": "openai-chat",
          "base_url": "https://api.synthetic.new",
          "path": "/openai/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "anthropic",
          "base_url": "https://api.synthetic.new",
          "path": "/anthropic/v1/messages",
          "protocol": "anthropic-messages"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming",
        "embeddings"
      ],
      "quirks": [
        {
          "text": "Prefer syn: routing aliases (syn:large:text etc.) — pinned model names 404 as models rotate.",
          "docs": "https://dev.synthetic.new/docs/api/models"
        },
        {
          "text": "Reasoning controls are not documented; behavior follows upstream chat templates — verify before relying.",
          "docs": "https://dev.synthetic.new/docs/api/models"
        }
      ],
      "plan": {
        "price_usd": 30,
        "period": "monthly",
        "quota": "$1/day or $30/month per pack, stackable; 500 price-weighted requests / 5h + $24/week credits + 1 concurrent per model",
        "notes": "Subscription includes all always-on models, no per-token billing; embeddings free and exempt.",
        "docs": "https://synthetic.new/rate-limits"
      },
      "offerings": [
        {
          "model": "alibaba/qwen3-8-27b",
          "wire_id": "hf:Qwen/Qwen3.8-27B",
          "endpoint": "openai-chat",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning controls are not documented; behavior follows the upstream Qwen chat template — verify before relying.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://api.synthetic.new/openai/v1/models",
              "verified": "2026-08-21"
            }
          }
        },
        {
          "model": "alibaba/qwen3-8-27b",
          "wire_id": "syn:small:vision",
          "endpoint": "openai-chat",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://dev.synthetic.new/docs/api/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Permanent category alias that routes to the current recommended upstream open-weight model (rotation possible). reasoning_effort low/medium/high documented; default undocumented — verify. Subscription-priced ($1/day or $30/mo, 500 requests/5h).",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://dev.synthetic.new/docs/openai/chat-completions",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "moonshot/kimi-k3",
          "wire_id": "hf:moonshotai/Kimi-K3",
          "endpoint": "openai-chat",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 524288,
            "output": null,
            "source": {
              "url": "https://dev.synthetic.new/docs/api/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning controls are not documented; behavior follows the upstream Kimi chat template — verify before relying. Catalog labels the model Beta.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://dev.synthetic.new/docs/api/models",
              "verified": "2026-08-19"
            }
          }
        },
        {
          "model": "moonshot/kimi-k3",
          "wire_id": "syn:large:vision",
          "endpoint": "openai-chat",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 524288,
            "output": null,
            "source": {
              "url": "https://dev.synthetic.new/docs/api/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Permanent category alias that routes to the current recommended upstream open-weight model (rotation possible). reasoning_effort low/medium/high documented; default undocumented — verify. Subscription-priced ($1/day or $30/mo, 500 requests/5h).",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://dev.synthetic.new/docs/openai/chat-completions",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "nvidia/nemotron-3-super",
          "wire_id": "hf:nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4",
          "endpoint": "openai-chat",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 262144,
            "output": null,
            "source": {
              "url": "https://dev.synthetic.new/docs/api/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Pinned to the Nemotron-3-Super NVFP4 build; upstream toggles thinking via chat_template_kwargs.enable_thinking (default on) — verify on this host.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://dev.synthetic.new/docs/api/models",
              "verified": "2026-08-28"
            },
            "toggle": {
              "param": "chat_template_kwargs.enable_thinking",
              "on": "true",
              "off": "false"
            }
          }
        },
        {
          "model": "openai/gpt-oss-120b",
          "wire_id": "hf:openai/gpt-oss-120b",
          "endpoint": "openai-chat",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 131072,
            "output": null,
            "source": {
              "url": "https://dev.synthetic.new/docs/api/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning controls are not documented; gpt-oss reasons by default upstream — verify before relying.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://dev.synthetic.new/docs/api/models",
              "verified": "2026-08-19"
            }
          }
        },
        {
          "model": "zai/glm-4-7-flash",
          "wire_id": "hf:zai-org/GLM-4.7-Flash",
          "endpoint": "openai-chat",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 196608,
            "output": null,
            "source": {
              "url": "https://dev.synthetic.new/docs/api/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Pinned to the GLM-4.7-Flash open weights (MIT). reasoning_effort low/medium/high documented; default undocumented — verify. Subscription-priced.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://dev.synthetic.new/docs/openai/chat-completions",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "zai/glm-4-7-flash",
          "wire_id": "syn:small:text",
          "endpoint": "openai-chat",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 196608,
            "output": null,
            "source": {
              "url": "https://dev.synthetic.new/docs/api/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Permanent category alias routing to the current recommended small text model (currently GLM-4.7-Flash; rotation possible). reasoning_effort low/medium/high documented; default undocumented — verify. Subscription-priced.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://dev.synthetic.new/docs/openai/chat-completions",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "zai/glm-5-2",
          "wire_id": "hf:zai-org/GLM-5.2",
          "endpoint": "openai-chat",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 524288,
            "output": null,
            "source": {
              "url": "https://dev.synthetic.new/docs/api/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Reasoning controls are not documented; behavior follows the upstream GLM chat template — verify before relying.",
            "returns": "hidden",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://dev.synthetic.new/docs/api/models",
              "verified": "2026-08-19"
            }
          }
        },
        {
          "model": "zai/glm-5-2",
          "wire_id": "syn:large:text",
          "endpoint": "openai-chat",
          "status": "ga",
          "status_date": "",
          "limits": {
            "context": 524288,
            "output": null,
            "source": {
              "url": "https://dev.synthetic.new/docs/api/models",
              "verified": "2026-08-28"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": false,
            "default": "on",
            "notes": "Permanent category alias that routes to the current recommended upstream open-weight model (rotation possible). reasoning_effort low/medium/high documented; default undocumented — verify. Subscription-priced ($1/day or $30/mo, 500 requests/5h).",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://dev.synthetic.new/docs/openai/chat-completions",
              "verified": "2026-08-28"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        }
      ]
    },
    {
      "id": "xai",
      "name": "xAI",
      "kind": "first_party",
      "urls": {
        "docs": "https://docs.x.ai/overview",
        "console": "https://console.x.ai"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "XAI_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create a key at console.x.ai. OpenAI SDKs work by changing base_url only; also Anthropic-compatible.",
          "docs": "https://docs.x.ai/overview"
        },
        {
          "id": "oauth",
          "type": "oauth",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [],
          "extra_headers": {},
          "getting_credentials": "grok login (or grok login --device-auth) with a SuperGrok or X Premium subscription. Tokens persist in ~/.grok/auth.json with background refresh; requests use subscription quota (a weekly pool shared across Grok chat, Build, and API), not API billing.",
          "docs": "https://docs.x.ai/build/overview",
          "flow": "browser OIDC or device code at auth.x.ai (endpoints not publicly documented)"
        }
      ],
      "endpoints": [
        {
          "id": "v1-chat-completions",
          "base_url": "https://api.x.ai",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "v1-responses",
          "base_url": "https://api.x.ai",
          "path": "/v1/responses",
          "protocol": "openai-responses"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Responses is the preferred API per xAI docs. presence_penalty, frequency_penalty, and stop cannot be used with reasoning models — requests including them error. grok-4.20-multi-agent maps reasoning.effort to agent count (4 or 16), not depth.",
          "docs": "https://docs.x.ai/docs/guides/reasoning"
        },
        {
          "text": "OAuth sessions call https://api.x.ai/v1 with Bearer tokens; an entitlement-aware catalog is available at https://cli-chat-proxy.grok.com/v1/models-v2. xAI decides which accounts receive OAuth tokens — some non-Heavy tiers report 403. SuperGrok is $30/mo, SuperGrok Plus $100/mo.",
          "docs": "https://docs.x.ai/build/enterprise"
        }
      ],
      "offerings": [
        {
          "model": "xai/grok-4-5",
          "wire_id": "grok-4.5",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 6,
            "cache_read": 0.3,
            "cache_write": null,
            "free": false,
            "notes": "Doubles for prompts over 200k.",
            "source": {
              "url": "https://docs.x.ai/docs/models",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 500000,
            "output": null,
            "source": {
              "url": "https://docs.x.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "notes": "xhigh is silently treated as high on 4.5.",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.x.ai/docs/guides/reasoning",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "xai/grok-4-5",
          "wire_id": "grok-4.5",
          "endpoint": "v1-responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 6,
            "cache_read": 0.3,
            "cache_write": null,
            "free": false,
            "notes": "Doubles for prompts over 200k.",
            "source": {
              "url": "https://docs.x.ai/docs/models",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 500000,
            "output": null,
            "source": {
              "url": "https://docs.x.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "hidden",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.x.ai/docs/guides/reasoning",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "xai/grok-4-6",
          "wire_id": "grok-4.6",
          "endpoint": "v1-chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 6,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "notes": "Doubles for prompts over 200k.",
            "source": {
              "url": "https://docs.x.ai/docs/models",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 500000,
            "output": null,
            "source": {
              "url": "https://docs.x.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.x.ai/docs/guides/reasoning",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning_effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "high"
            }
          }
        },
        {
          "model": "xai/grok-4-6",
          "wire_id": "grok-4.6",
          "endpoint": "v1-responses",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 2,
            "output": 6,
            "cache_read": 0.5,
            "cache_write": null,
            "free": false,
            "notes": "Doubles for prompts over 200k.",
            "source": {
              "url": "https://docs.x.ai/docs/models",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 500000,
            "output": null,
            "source": {
              "url": "https://docs.x.ai/docs/models",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "effort",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "encrypted_content",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.x.ai/docs/guides/reasoning",
              "verified": "2026-08-18"
            },
            "effort": {
              "param": "reasoning.effort",
              "values": [
                "low",
                "medium",
                "high",
                "xhigh"
              ],
              "default": "high"
            }
          }
        }
      ]
    },
    {
      "id": "xiaomi-mimo",
      "name": "Xiaomi MiMo",
      "kind": "first_party",
      "urls": {
        "docs": "https://mimo.mi.com/docs/en-US/quickstart",
        "console": "https://platform.xiaomimimo.com"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "MIMO_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "Create a key at platform.xiaomimimo.com; the api-key: header is also accepted.",
          "docs": "https://mimo.mi.com/docs/en-US/quickstart"
        }
      ],
      "endpoints": [
        {
          "id": "chat-completions",
          "base_url": "https://api.xiaomimimo.com",
          "path": "/v1/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "messages",
          "base_url": "https://api.xiaomimimo.com",
          "path": "/v1/messages",
          "protocol": "anthropic-messages"
        },
        {
          "id": "responses",
          "base_url": "https://api.xiaomimimo.com",
          "path": "/v1/responses",
          "protocol": "openai-responses"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Auth accepts api-key: header or Authorization: Bearer.",
          "docs": "https://mimo.mi.com/docs/en-US/quickstart"
        },
        {
          "text": "top_p ignored in thinking mode; reasoning returns in reasoning_content and must be passed back in tool loops.",
          "docs": "https://mimo.mi.com/docs/en-US/deep-thinking"
        },
        {
          "text": "Token Plan subscriptions also offered; V2 series retired 2026-06-30; ultraspeed FP4 variant early access.",
          "docs": "https://mimo.mi.com/docs/en-US/quickstart"
        }
      ],
      "offerings": [
        {
          "model": "xiaomi/mimo-v2-5",
          "wire_id": "mimo-v2.5",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.14,
            "output": 0.28,
            "cache_read": 0.0028,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://mimo.mi.com/docs/en-US/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://mimo.mi.com/docs/en-US/quickstart",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Deep Thinking mode; toggle parameter not documented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://mimo.mi.com/docs/en-US/deep-thinking",
              "verified": "2026-08-20"
            }
          }
        },
        {
          "model": "xiaomi/mimo-v2-5-pro",
          "wire_id": "mimo-v2.5-pro",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.435,
            "output": 0.87,
            "cache_read": 0.0036,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://mimo.mi.com/docs/en-US/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://mimo.mi.com/docs/en-US/quickstart",
              "verified": "2026-08-20"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "adaptive",
            "mandatory": false,
            "default": "on",
            "notes": "Deep Thinking mode; toggle parameter not documented — verify.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://mimo.mi.com/docs/en-US/deep-thinking",
              "verified": "2026-08-20"
            }
          }
        }
      ]
    },
    {
      "id": "zai",
      "name": "Z.ai",
      "kind": "first_party",
      "urls": {
        "docs": "https://docs.z.ai/guides/llm/glm-5.3",
        "console": "https://z.ai/manage-apikey/apikey-list",
        "pricing": "https://docs.z.ai/guides/develop/http/introduction"
      },
      "auth": [
        {
          "id": "api-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "ZAI_API_KEY"
          ],
          "extra_headers": {},
          "getting_credentials": "z.ai console → API Keys. Key format id.secret; JWT HS256 derivation also supported.",
          "docs": "https://docs.z.ai/guides/develop/http/introduction"
        }
      ],
      "endpoints": [
        {
          "id": "chat-completions",
          "base_url": "https://api.z.ai",
          "path": "/api/paas/v4/chat/completions",
          "protocol": "openai-chat"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "clear_thinking: false preserves reasoning_content across turns (Preserved Thinking); disabled by default on this endpoint.",
          "docs": "https://docs.z.ai/guides/capabilities/thinking-mode"
        },
        {
          "text": "GLM-5.3-Flash is the first multimodal GLM-5-series model: chat-completions accepts image/video/file inputs alongside text.",
          "docs": "https://docs.z.ai/guides/vlm/glm-5.3-flash"
        }
      ],
      "offerings": [
        {
          "model": "zai/glm-4-6",
          "wire_id": "glm-4.6",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.6,
            "output": 2.2,
            "cache_read": 0.11,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://docs.z.ai/guides/overview/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 200000,
            "output": null,
            "source": {
              "url": "https://docs.z.ai/guides/llm/glm-4.6",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Hybrid thinking on by default.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.z.ai/guides/llm/glm-4.6",
              "verified": "2026-08-18"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "zai/glm-5-1",
          "wire_id": "glm-5.1",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.4,
            "output": 4.4,
            "cache_read": 0.26,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://docs.z.ai/guides/llm/glm-5.1",
              "verified": "2026-08-21"
            }
          },
          "limits": {
            "context": 204800,
            "output": 131072,
            "source": {
              "url": "https://docs.z.ai/guides/llm/glm-5.1",
              "verified": "2026-08-21"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Thinking default enabled; thinking.type toggles.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.z.ai/guides/llm/glm-5.1",
              "verified": "2026-08-21"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "zai/glm-5-3",
          "wire_id": "glm-5.3",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 1.4,
            "output": 4.4,
            "cache_read": 0.26,
            "cache_write": null,
            "free": false,
            "source": {
              "url": "https://docs.z.ai/guides/overview/pricing",
              "verified": "2026-08-20"
            }
          },
          "limits": {
            "context": 1048576,
            "output": 128000,
            "source": {
              "url": "https://docs.z.ai/guides/llm/glm-5.3",
              "verified": "2026-08-19"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "notes": "Forced thinking: thinking.type enabled and disabled both return 400.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.z.ai/guides/llm/glm-5.3",
              "verified": "2026-08-18"
            }
          }
        },
        {
          "model": "zai/glm-5-3-flash",
          "wire_id": "glm-5.3-flash",
          "endpoint": "chat-completions",
          "status": "ga",
          "status_date": "",
          "cost": {
            "input": 0.075,
            "output": 0.25,
            "cache_read": 0.015,
            "cache_write": null,
            "free": false,
            "notes": "Launch promo 50% off until Sep 9 2026 UTC+8; regular rates $0.15/$0.50/$0.03.",
            "source": {
              "url": "https://docs.z.ai/guides/overview/pricing",
              "verified": "2026-08-26"
            }
          },
          "limits": {
            "context": 1048576,
            "output": null,
            "source": {
              "url": "https://docs.z.ai/guides/vlm/glm-5.3-flash",
              "verified": "2026-08-26"
            }
          },
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.z.ai/guides/vlm/glm-5.3-flash",
              "verified": "2026-08-26"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": ""
            }
          }
        }
      ]
    },
    {
      "id": "zai-coding-plan",
      "name": "Z.ai GLM Coding Plan",
      "kind": "subscription",
      "urls": {
        "docs": "https://docs.z.ai/devpack/overview",
        "console": "https://z.ai/subscribe"
      },
      "auth": [
        {
          "id": "plan-key",
          "type": "api_key",
          "transport": "header",
          "header": "Authorization: Bearer",
          "env": [
            "ZAI_CODING_PLAN_API_KEY"
          ],
          "key_prefix": "",
          "extra_headers": {},
          "getting_credentials": "Subscribe at z.ai/subscribe; plan keys are distinct from normal API keys and only work on coding-plan endpoints.",
          "docs": "https://docs.z.ai/devpack/quick-start"
        }
      ],
      "endpoints": [
        {
          "id": "coding-chat-completions",
          "base_url": "https://api.z.ai",
          "path": "/api/coding/paas/v4/chat/completions",
          "protocol": "openai-chat"
        },
        {
          "id": "coding-anthropic",
          "base_url": "https://api.z.ai",
          "path": "/api/anthropic",
          "protocol": "anthropic-messages"
        },
        {
          "id": "coding-responses",
          "base_url": "https://api.z.ai",
          "path": "/api/v1",
          "protocol": "openai-responses"
        }
      ],
      "api_surfaces": [
        "text",
        "streaming"
      ],
      "quirks": [
        {
          "text": "Preserved Thinking is enabled by default on coding-plan endpoints (opposite of the standard API). For Claude Code: ANTHROPIC_BASE_URL=https://api.z.ai/api/anthropic, ANTHROPIC_AUTH_TOKEN=<plan key>.",
          "docs": "https://docs.z.ai/devpack/quick-start"
        }
      ],
      "plan": {
        "price_usd": 18,
        "period": "monthly",
        "quota": "Lite ~2000 credits/5h, 10000/week; Pro ~12000/5h; Max ~28000/5h",
        "notes": "Credits-based since 2026-07-30. Includes GLM-5.3, GLM-5-Turbo, GLM-4.7; GLM-5.2/5.1 auto-route to GLM-5.3.",
        "docs": "https://docs.z.ai/devpack/overview"
      },
      "offerings": [
        {
          "model": "zai/glm-4-6",
          "wire_id": "glm-4.6",
          "endpoint": "coding-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "toggle",
            "mandatory": false,
            "default": "on",
            "notes": "Preserved Thinking on by default on this endpoint.",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.z.ai/guides/capabilities/thinking-mode",
              "verified": "2026-08-18"
            },
            "toggle": {
              "param": "thinking.type",
              "on": "enabled",
              "off": "disabled"
            }
          }
        },
        {
          "model": "zai/glm-5-3",
          "wire_id": "glm-5.3",
          "endpoint": "coding-chat-completions",
          "status": "ga",
          "status_date": "",
          "features": {
            "streaming": true,
            "tools": true,
            "structured_output": true,
            "prompt_caching": true,
            "vision": true
          },
          "reasoning": {
            "style": "always_on",
            "mandatory": true,
            "default": "on",
            "returns": "reasoning_content",
            "must_round_trip": "reasoning_content",
            "incompatible_with": [],
            "source": {
              "url": "https://docs.z.ai/devpack/overview",
              "verified": "2026-08-18"
            }
          }
        }
      ]
    }
  ],
  "models": [
    {
      "id": "alibaba/qwen3-7-flash",
      "name": "Qwen3.7 Flash",
      "family": "qwen3",
      "lab": "alibaba",
      "release_date": "2026-07-27",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "qwen3.7-flash"
      ],
      "description": "Cheapest 1M-context vision tier: native text+image+video input, 131K output. $0.03/$0.13 per Mtok on OpenRouter (Alibaba tiered CNY 0.2/0.8 up to 32K in Beijing). Params undisclosed; API-only. A Thinking Mode exists but the control parameter is undocumented.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "qwen/qwen3.7-flash",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "alibaba/qwen3-7-max",
      "name": "Qwen3.7 Max",
      "family": "qwen3",
      "lab": "alibaba",
      "release_date": "2026-05-20",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "qwen3.7-max"
      ],
      "description": "Qwen's most capable proprietary model at the 3.7 generation (snapshot 2026-05-20); 1M context, 131K output, text input, thinking mode supported. CNY 12/36 per Mtok (Beijing); the 2026-06-08 snapshot adds vision.",
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "qwen/qwen3.7-max",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-go",
          "wire_id": "qwen3.7-max",
          "endpoint": "go-chat-completions"
        }
      ]
    },
    {
      "id": "alibaba/qwen3-7-plus",
      "name": "Qwen3.7 Plus",
      "family": "qwen3",
      "lab": "alibaba",
      "release_date": "2026-06-02",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "qwen3.7-plus"
      ],
      "description": "Qwen3.7 series; thinking ON by default; preserve_thinking carries reasoning_content across turns. Announced June 1–3, 2026 (sources differ on exact day).",
      "ranking": {
        "aa_index": 39,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/qwen3-7-plus",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "alibaba-dashscope",
          "wire_id": "qwen3.7-plus",
          "endpoint": "anthropic"
        },
        {
          "provider": "alibaba-dashscope",
          "wire_id": "qwen3.7-plus",
          "endpoint": "compatible-mode-chat"
        },
        {
          "provider": "opencode-go",
          "wire_id": "qwen3.7-plus",
          "endpoint": "go-messages"
        },
        {
          "provider": "openrouter",
          "wire_id": "qwen/qwen3.7-plus",
          "endpoint": "chat-completions"
        },
        {
          "provider": "qwen-coding-plan",
          "wire_id": "qwen3.7-plus",
          "endpoint": "coding-chat"
        }
      ]
    },
    {
      "id": "alibaba/qwen3-8-2-4t-a95b",
      "name": "Qwen3.8 2.4T A95B",
      "family": "qwen3",
      "lab": "alibaba",
      "release_date": "2026-08-12",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "Qwen/Qwen3.8-2.4T-A95B",
      "license": "Qwen3.8-Max (custom)",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "qwen3.8-2.4t-a95b",
        "qwen3.8-2.4t"
      ],
      "description": "Open-weight Max-class flagship: 2.4T total / 95B active MoE (512 experts, hybrid Gated DeltaNet/Attention), 262K native context extensible past 1M, 131K output. Thinking is mandatory (enable_thinking=false errors); reasoning_effort xhigh (default)/medium/low, preserve_thinking on. $2.00/$6.00 per Mtok.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "qwen/qwen3.8-2.4t-a95b",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "alibaba/qwen3-8-27b",
      "name": "Qwen3.8 27B",
      "family": "qwen3",
      "lab": "alibaba",
      "release_date": "2026-08-14",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "Qwen/Qwen3.8-27B"
      ],
      "description": "Dense 27B with hybrid thinking, 262K context, vision input. Release date per third-party reporting.",
      "ranking": {
        "aa_index": 52,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/qwen3-8-27b",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "hetzner",
          "wire_id": "Qwen/Qwen3.8-27B",
          "endpoint": "chat"
        },
        {
          "provider": "io-intelligence",
          "wire_id": "Qwen/Qwen3.8-27B",
          "endpoint": "chat"
        },
        {
          "provider": "near-ai",
          "wire_id": "Qwen/Qwen3.8-27B",
          "endpoint": "chat"
        },
        {
          "provider": "openrouter",
          "wire_id": "qwen/qwen3.8-27b",
          "endpoint": "chat-completions"
        },
        {
          "provider": "synthetic",
          "wire_id": "hf:Qwen/Qwen3.8-27B",
          "endpoint": "openai-chat"
        },
        {
          "provider": "synthetic",
          "wire_id": "syn:small:vision",
          "endpoint": "openai-chat"
        }
      ]
    },
    {
      "id": "alibaba/qwen3-8-flash",
      "name": "Qwen3.8 Flash",
      "family": "qwen",
      "lab": "alibaba",
      "release_date": "2026-08-26",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "Qwen/Qwen3.8-Flash-Next",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "qwen3.8-flash",
        "Qwen3.8-Flash"
      ],
      "description": "Production vision-language MoE (125B LM params / 6B active); 262K context extensible to 1M via YaRN; API coming soon on QwenCloud — preview architecture for Qwen4.",
      "offered_via": [
        {
          "provider": "opencode-go",
          "wire_id": "qwen3.8-flash",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "qwen/qwen3.8-flash",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "alibaba/qwen3-8-max",
      "name": "Qwen3.8 Max",
      "family": "qwen3",
      "lab": "alibaba",
      "release_date": "2026-08-03",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "qwen3.8-max"
      ],
      "description": "Flagship MoE (~2.4T), 1M context; $2/$6 per Mtok; first Max-class promised open — weights pending. Preview 2026-07-19.",
      "ranking": {
        "aa_index": 58,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/qwen3-8-max",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "alibaba-dashscope",
          "wire_id": "qwen3.8-max",
          "endpoint": "compatible-mode-chat"
        },
        {
          "provider": "opencode-go",
          "wire_id": "qwen3.8-max",
          "endpoint": "go-messages"
        },
        {
          "provider": "openrouter",
          "wire_id": "qwen/qwen3.8-max-0902",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "allenai/olmo-3-32b-think",
      "name": "OLMo 3 32B Think",
      "family": "olmo",
      "lab": "allenai",
      "release_date": "2025-11-20",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "allenai/Olmo-3-32B-Think",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "olmo-3-32b-think"
      ],
      "description": "Fully open (Apache 2.0, all data + code + logs) 32B dense long-CoT reasoning model; 64K context.",
      "offered_via": [
        {
          "provider": "opencode-zen",
          "wire_id": "olmo-3-32b-think",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "allenai/olmo-3-32b-think",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "amazon/nova-2-lite",
      "name": "Nova 2 Lite",
      "family": "nova",
      "lab": "amazon",
      "release_date": "2025-12-02",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "nova-2-lite-v1"
      ],
      "description": "AWS re:Invent 2025 next-gen Nova; 1M context, text+image+video+file input. Reasoning via Bedrock reasoningConfig (type enabled/disabled + maxReasoningEffort low/medium/high, default disabled). $0.30/$2.50 per Mtok; more capable than Nova Premier at ~7x lower cost.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "amazon/nova-2-lite-v1",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "anthropic/claude-fable-5-1",
      "name": "Claude Fable 5.1",
      "family": "claude",
      "lab": "anthropic",
      "release_date": "2026-09-01",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "claude-fable-5.1"
      ],
      "description": "GA refresh of Fable 5 (Sep 1, 2026): same $10/$50 per Mtok with prompt-cache hits cut from $1.00 to $0.25; 1M context, 128K output, text+image input. Fable line still cannot disable thinking.",
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "anthropic/claude-fable-5-1",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "claude-fable-5-1",
          "endpoint": "messages"
        },
        {
          "provider": "openrouter",
          "wire_id": "anthropic/claude-fable-5.1",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "anthropic/claude-fable-5",
      "name": "Claude Fable 5",
      "family": "claude",
      "lab": "anthropic",
      "release_date": "2026-06-09",
      "retired_date": "",
      "knowledge_cutoff": "2026-01-31",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Top tier for long-horizon reasoning (2x Opus pricing); thinking cannot be disabled; mandatory 30-day retention; refusals return stop_reason refusal.",
      "ranking": {
        "aa_index": 62,
        "aa_variant": "Adaptive Reasoning, Max Effort",
        "url": "https://artificialanalysis.ai/models/claude-fable-5",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "anthropic",
          "wire_id": "claude-fable-5",
          "endpoint": "v1-messages"
        },
        {
          "provider": "aws-bedrock",
          "wire_id": "anthropic.claude-fable-5",
          "endpoint": "invoke-anthropic"
        },
        {
          "provider": "near-ai",
          "wire_id": "anthropic/claude-fable-5",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "claude-fable-5",
          "endpoint": "messages"
        },
        {
          "provider": "openrouter",
          "wire_id": "anthropic/claude-fable-5",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "anthropic/claude-haiku-4-5",
      "name": "Claude Haiku 4.5",
      "family": "claude",
      "lab": "anthropic",
      "release_date": "2025-10-01",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "claude-haiku-4-5-20251001"
      ],
      "description": "Fast Claude; supports extended thinking but not interleaved thinking.",
      "ranking": {
        "aa_index": 30,
        "aa_variant": "reasoning",
        "url": "https://artificialanalysis.ai/models/claude-4-5-haiku-reasoning",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "anthropic",
          "wire_id": "claude-haiku-4-5",
          "endpoint": "v1-messages"
        },
        {
          "provider": "aws-bedrock",
          "wire_id": "anthropic.claude-haiku-4-5-20251001-v1:0",
          "endpoint": "invoke-anthropic"
        },
        {
          "provider": "near-ai",
          "wire_id": "anthropic/claude-haiku-4-5",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "claude-haiku-4-5",
          "endpoint": "messages"
        },
        {
          "provider": "openrouter",
          "wire_id": "anthropic/claude-haiku-4.5",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "anthropic/claude-opus-4-6",
      "name": "Claude Opus 4.6",
      "family": "claude",
      "lab": "anthropic",
      "release_date": "2026-02-04",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "claude-opus-4.6"
      ],
      "description": "February 2026 Opus release, $5/$25 per Mtok with 1M context and 128K output — same price as Opus 5; part of the 4.x line Anthropic ships alongside the 5-series.",
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "anthropic/claude-opus-4-6",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "claude-opus-4-6",
          "endpoint": "messages"
        },
        {
          "provider": "openrouter",
          "wire_id": "anthropic/claude-opus-4.6",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "anthropic/claude-opus-4-7",
      "name": "Claude Opus 4.7",
      "family": "claude",
      "lab": "anthropic",
      "release_date": "2026-04-16",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "claude-opus-4.7"
      ],
      "description": "April 2026 Opus release, $5/$25 per Mtok with 1M context and 128K output.",
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "anthropic/claude-opus-4-7",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "claude-opus-4-7",
          "endpoint": "messages"
        },
        {
          "provider": "openrouter",
          "wire_id": "anthropic/claude-opus-4.7",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "anthropic/claude-opus-4-8",
      "name": "Claude Opus 4.8",
      "family": "claude",
      "lab": "anthropic",
      "release_date": "2026-05-27",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "claude-opus-4.8",
        "claude-opus-4-8-fast"
      ],
      "description": "May 2026 Opus release and latest of the 4.x line; $5/$25 per Mtok standard with 1M context and 128K output, fast mode $10/$50, ~$10/$37.50 premium above 200K prompt tokens.",
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "anthropic/claude-opus-4-8",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "claude-opus-4-8",
          "endpoint": "messages"
        },
        {
          "provider": "openrouter",
          "wire_id": "anthropic/claude-opus-4.8-fast",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "anthropic/claude-opus-4.8",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "anthropic/claude-opus-5",
      "name": "Claude Opus 5",
      "family": "claude",
      "lab": "anthropic",
      "release_date": "2026-07-24",
      "retired_date": "",
      "knowledge_cutoff": "2026-05-31",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Frontier Opus; adaptive thinking (output_config.effort, default high); disabling accepted only at effort up to high.",
      "ranking": {
        "aa_index": 63,
        "aa_variant": "Adaptive Reasoning, Max Effort",
        "url": "https://artificialanalysis.ai/models/claude-opus-5",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "anthropic",
          "wire_id": "claude-opus-5",
          "endpoint": "v1-messages"
        },
        {
          "provider": "aws-bedrock",
          "wire_id": "anthropic.claude-opus-5",
          "endpoint": "invoke-anthropic"
        },
        {
          "provider": "near-ai",
          "wire_id": "anthropic/claude-opus-5",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "claude-opus-5",
          "endpoint": "messages"
        },
        {
          "provider": "openrouter",
          "wire_id": "anthropic/claude-opus-5",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "anthropic/claude-sonnet-5",
      "name": "Claude Sonnet 5",
      "family": "claude",
      "lab": "anthropic",
      "release_date": "2026-06-30",
      "retired_date": "",
      "knowledge_cutoff": "2026-01-31",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Mainstream 5-series; adaptive thinking; $2/$10 pricing made permanent 2026-08-10.",
      "ranking": {
        "aa_index": 55,
        "aa_variant": "Adaptive Reasoning, Max Effort",
        "url": "https://artificialanalysis.ai/models/claude-sonnet-5",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "anthropic",
          "wire_id": "claude-sonnet-5",
          "endpoint": "v1-messages"
        },
        {
          "provider": "aws-bedrock",
          "wire_id": "anthropic.claude-sonnet-5",
          "endpoint": "invoke-anthropic"
        },
        {
          "provider": "near-ai",
          "wire_id": "anthropic/claude-sonnet-5",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "claude-sonnet-5",
          "endpoint": "messages"
        },
        {
          "provider": "openrouter",
          "wire_id": "anthropic/claude-sonnet-5",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "baidu/ernie-4-5-vl",
      "name": "ERNIE 4.5 VL 424B A47B",
      "family": "ernie",
      "lab": "baidu",
      "release_date": "2025-06-30",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "baidu/ERNIE-4.5-VL-424B-A47B-PT",
      "license": "Apache-2.0",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "ernie-4.5-vl-424b-a47b"
      ],
      "description": "Baidu's largest open-weights ERNIE: heterogeneous 424B/47B-active MoE with modality-isolated routing, text+image input, 131K context. Chat-template enable_thinking (default on). Still the current open ERNIE flagship.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "baidu/ernie-4.5-vl-424b-a47b",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "bytedance/seed-2-0-code",
      "name": "Seed 2.0 Code",
      "family": "seed",
      "lab": "bytedance",
      "release_date": "2026-02-14",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "seed-2.0-code"
      ],
      "description": "Agentic-coding flagship of the Seed 2.0 launch (Pro/Lite/Mini + Code); 262K context, multimodal input, 131K output. Tiered $0.50/$3.00 per Mtok up to 128K prompt, 2x above. Ark thinking object enabled/disabled/auto plus reasoning_effort.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "bytedance-seed/seed-2.0-code",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "bytedance/seed-2-0-lite",
      "name": "Seed 2.0 Lite",
      "family": "seed",
      "lab": "bytedance",
      "release_date": "2026-02-14",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "seed-2.0-lite"
      ],
      "description": "Seed 2.0 Lite tier; 262K context, multimodal input, 131K output. Tiered $0.25/$2.00 per Mtok up to 128K prompt, 2x above; post-April 2026 update added native audio understanding.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "bytedance-seed/seed-2.0-lite",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "bytedance/seed-2-0-mini",
      "name": "Seed 2.0 Mini",
      "family": "seed",
      "lab": "bytedance",
      "release_date": "2026-02-14",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "seed-2.0-mini"
      ],
      "description": "Seed 2.0 latency/cost tier; 262K context, multimodal input, 131K output. Tiered $0.10/$0.40 per Mtok up to 128K prompt, 2x above. Effort modes minimal/low/medium/high.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "bytedance-seed/seed-2.0-mini",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "bytedance/seed-2-1-turbo",
      "name": "Seed 2.1 Turbo",
      "family": "seed",
      "lab": "bytedance",
      "release_date": "2026-06-23",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "seed-2.1-turbo"
      ],
      "description": "Newest Seed generation (Volcengine FORCE, June 23-24 2026), positioned for coding and long-horizon agents; 262K context, multimodal input, up to 262K output (235,929 enforced). $0.50/$2.50 per Mtok. Ark thinking object enabled/disabled/auto.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "bytedance-seed/seed-2-1-turbo",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "cohere/command-a-plus",
      "name": "Command A+ (05-2026)",
      "family": "command",
      "lab": "cohere",
      "release_date": "2026-05-20",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "",
      "license": "Apache-2.0",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "command-a-plus-05-2026"
      ],
      "description": "218B/25B MoE, vision, 48 languages, Apache 2.0; free until rate limits on Cohere's platform.",
      "ranking": {
        "aa_index": 23,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/command-a-plus",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "cohere",
          "wire_id": "command-a-plus-05-2026",
          "endpoint": "compat-chat-completions"
        }
      ]
    },
    {
      "id": "cohere/command-a-reasoning",
      "name": "Command A Reasoning (08-2025)",
      "family": "command",
      "lab": "cohere",
      "release_date": "2025-08-21",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "CohereLabs/command-a-reasoning-08-2025",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "command-a-reasoning-08-2025"
      ],
      "description": "111B dense, Cohere's first reasoning model; native thinking parameter with budget_tokens.",
      "offered_via": [
        {
          "provider": "cohere",
          "wire_id": "command-a-reasoning-08-2025",
          "endpoint": "compat-chat-completions"
        }
      ]
    },
    {
      "id": "deepseek/deepseek-v4-flash",
      "name": "DeepSeek V4 Flash",
      "family": "deepseek-v4",
      "lab": "deepseek",
      "release_date": "2026-07-31",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "deepseek-v4-flash-0731"
      ],
      "description": "Hybrid thinking on by default (V4-Flash-0731); 1M context, 384K max output shared with CoT.",
      "ranking": {
        "aa_index": 52,
        "aa_variant": "V4 Flash 0731, max",
        "url": "https://artificialanalysis.ai/models/deepseek-v4-flash",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "baseten",
          "wire_id": "deepseek-ai/DeepSeek-V4-Flash-0731",
          "endpoint": "chat-completions"
        },
        {
          "provider": "deepseek",
          "wire_id": "deepseek-v4-flash",
          "endpoint": "anthropic"
        },
        {
          "provider": "deepseek",
          "wire_id": "deepseek-v4-flash",
          "endpoint": "chat-completions"
        },
        {
          "provider": "io-intelligence",
          "wire_id": "deepseek-ai/DeepSeek-V4-Flash",
          "endpoint": "chat"
        },
        {
          "provider": "near-ai",
          "wire_id": "deepseek-ai/DeepSeek-V4-Flash",
          "endpoint": "chat"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "deepseek-v4-flash:0731",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "deepseek-v4-flash",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "deepseek-v4-flash",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "deepseek/deepseek-v4-flash-0731",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "deepseek/deepseek-v4-flash",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "deepseek/deepseek-v4-pro",
      "name": "DeepSeek V4 Pro",
      "family": "deepseek-v4",
      "lab": "deepseek",
      "release_date": "2026-08-13",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "deepseek-v4-pro-0813"
      ],
      "description": "Hybrid thinking on by default (V4-Pro-0813); 1M context, 384K max output shared with CoT.",
      "ranking": {
        "aa_index": 53,
        "aa_variant": "V4 Pro 0813, max effort",
        "url": "https://artificialanalysis.ai/models/deepseek-v4-pro",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "alibaba-token-plan",
          "wire_id": "deepseek-v4-pro",
          "endpoint": "compatible-mode"
        },
        {
          "provider": "deepseek",
          "wire_id": "deepseek-v4-pro",
          "endpoint": "anthropic"
        },
        {
          "provider": "deepseek",
          "wire_id": "deepseek-v4-pro",
          "endpoint": "chat-completions"
        },
        {
          "provider": "fireworks-ai",
          "wire_id": "deepseek-v4-pro",
          "endpoint": "inference-v1"
        },
        {
          "provider": "io-intelligence",
          "wire_id": "deepseek-ai/DeepSeek-V4-Pro",
          "endpoint": "chat"
        },
        {
          "provider": "nvidia",
          "wire_id": "deepseek-ai/deepseek-v4-pro",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "deepseek-v4-pro:0813",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "deepseek-v4-pro",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "deepseek-v4-pro",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "deepseek/deepseek-v4-pro-0813",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "deepseek/deepseek-v4-pro",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "google/gemini-3-1-pro",
      "name": "Gemini 3.1 Pro",
      "family": "gemini",
      "lab": "google",
      "release_date": "2026-02-19",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "gemini-3.1-pro-preview",
        "gemini-3-pro-preview"
      ],
      "description": "Preview-only; tiered pricing (≤200K $2/$12, >200K $4/$18); carries the gemini-3-pro-preview alias since 2026-03-09.",
      "offered_via": [
        {
          "provider": "google-gemini",
          "wire_id": "gemini-3.1-pro-preview",
          "endpoint": "generate-content"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gemini-3.1-pro",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "google/gemini-3.1-pro-preview",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "google/gemini-3-5-flash-lite",
      "name": "Gemini 3.5 Flash-Lite",
      "family": "gemini",
      "lab": "google",
      "release_date": "2026-07-21",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Cheapest tier; default minimal — docs recommend medium/high for multi-step subagent work; temperature/top_p/top_k deprecated (3.6/3.7).",
      "ranking": {
        "aa_index": 37,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/gemini-3-5-flash-lite",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "google-gemini",
          "wire_id": "gemini-3.5-flash-lite",
          "endpoint": "generate-content"
        },
        {
          "provider": "google-vertex",
          "wire_id": "gemini-3.5-flash-lite",
          "endpoint": "generate-content"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gemini-3.5-flash-lite",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "google/gemini-3.5-flash-lite",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "google/gemini-3-5-flash",
      "name": "Gemini 3.5 Flash",
      "family": "gemini",
      "lab": "google",
      "release_date": "2026-05-19",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "video",
          "audio"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "gemini-3.5-flash"
      ],
      "description": "GA Flash generation (migrated from gemini-3-flash-preview); 1M context, 65K output, text+image+video+audio+PDF input. $1.50/$9.00 per Mtok flat (output price includes thinking); caching $0.15. Controlled via thinking_level minimal/low/medium/high, default medium; thinking_budget is deprecated and sending both returns 400.",
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "google/gemini-3.5-flash",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gemini-3.5-flash",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "google/gemini-3.5-flash",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "google/gemini-3-6-flash",
      "name": "Gemini 3.6 Flash",
      "family": "gemini",
      "lab": "google",
      "release_date": "2026-07-21",
      "retired_date": "",
      "knowledge_cutoff": "2026-03-31",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Intro pricing (50% off) through 2026-12-31.",
      "ranking": {
        "aa_index": 52,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/gemini-3-6-flash",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "google-gemini",
          "wire_id": "gemini-3.6-flash",
          "endpoint": "generate-content"
        },
        {
          "provider": "google-vertex",
          "wire_id": "gemini-3.6-flash",
          "endpoint": "generate-content"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gemini-3.6-flash",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "google/gemini-3.6-flash",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "google/gemini-3-7-flash",
      "name": "Gemini 3.7 Flash",
      "family": "gemini",
      "lab": "google",
      "release_date": "2026-08-13",
      "retired_date": "",
      "knowledge_cutoff": "2026-03-31",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Flash flagship; minimal thinking level returns an error.",
      "ranking": {
        "aa_index": 56,
        "aa_variant": "high effort",
        "url": "https://artificialanalysis.ai/models/gemini-3-7-flash",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "google-gemini",
          "wire_id": "gemini-3.7-flash",
          "endpoint": "generate-content"
        },
        {
          "provider": "google-vertex",
          "wire_id": "gemini-3.7-flash",
          "endpoint": "generate-content"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gemini-3.7-flash",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "google/gemini-3.7-flash",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "google/gemini-3-8-flash",
      "name": "Gemini 3.8 Flash",
      "family": "gemini",
      "lab": "google",
      "release_date": "2026-09-03",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "audio",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "gemini-3.8-flash"
      ],
      "description": "Fourth Flash generation (Sep 3, 2026): 1M context, text+image+audio+video input, $0.75/$3.75 per Mtok on OpenRouter (Google list $1.50/$7.50, cache $0.15). thinking_level control like the 3.5+ Flash line.",
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "google/gemini-3.8-flash",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gemini-3.8-flash",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "google/gemini-3.8-flash",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "google/gemini-3-flash",
      "name": "Gemini 3 Flash",
      "family": "gemini",
      "lab": "google",
      "release_date": "2025-12-17",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "gemini-3-flash-preview"
      ],
      "description": "Preview-only on the Gemini API (gemini-3-flash-preview); superseded by 3.5+ Flash GA generations.",
      "offered_via": [
        {
          "provider": "google-gemini",
          "wire_id": "gemini-3-flash-preview",
          "endpoint": "generate-content"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gemini-3-flash",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "google/gemini-3-flash-preview",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "google/gemini-3-pro",
      "name": "Gemini 3 Pro",
      "family": "gemini",
      "lab": "google",
      "release_date": "2025-11-18",
      "retired_date": "",
      "knowledge_cutoff": "2025-01-31",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "gemini-3-pro-preview"
      ],
      "description": "Uses thinkingLevel (low/high); cannot disable thinking.",
      "ranking": {
        "aa_index": 41,
        "aa_variant": "Gemini 3 Pro Preview (high)",
        "url": "https://artificialanalysis.ai/models/gemini-3-pro",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "google-gemini",
          "wire_id": "gemini-3-pro-preview",
          "endpoint": "generate-content"
        },
        {
          "provider": "google-vertex",
          "wire_id": "gemini-3-pro-preview",
          "endpoint": "generate-content"
        }
      ]
    },
    {
      "id": "google/gemma-4",
      "name": "Gemma 4",
      "family": "gemma",
      "lab": "google",
      "release_date": "2026-03-31",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "audio",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "gemma-4-31B-it",
        "gemma4"
      ],
      "description": "Open-weight multimodal family (E2B/E4B/26B-A4B/31B dense), 256K context; MTP variants 2026-04-16, 12B Unified 2026-06-03; now served on the Gemini API.",
      "ranking": {
        "aa_index": 30,
        "aa_variant": "Gemma 4 31B reasoning",
        "url": "https://artificialanalysis.ai/models/gemma-4-31b",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "google-gemini",
          "wire_id": "gemma-4-31b-it",
          "endpoint": "generate-content"
        },
        {
          "provider": "io-intelligence",
          "wire_id": "google/gemma-4-26b-a4b-it",
          "endpoint": "chat"
        },
        {
          "provider": "near-ai",
          "wire_id": "google/gemma-4-31B-it",
          "endpoint": "chat"
        },
        {
          "provider": "nvidia",
          "wire_id": "google/gemma-4-31b-it",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "gemma4:31b",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "google/gemma-4-26b-a4b-it",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "google/gemma-4-31b-it",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "ibm/granite-4-1-8b",
      "name": "Granite 4.1 8B",
      "family": "granite",
      "lab": "ibm",
      "release_date": "2026-04-29",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "ibm-granite/granite-4.1-8b",
      "license": "Apache-2.0",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "granite-4.1-8b"
      ],
      "description": "IBM's current small long-context instruct model: 8B dense, 131K context, Apache-2.0. $0.05/$0.10 per Mtok on OpenRouter.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "ibm-granite/granite-4.1-8b",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "ibm/granite-4-2-8b",
      "name": "Granite 4.2 8B",
      "family": "granite",
      "lab": "ibm",
      "release_date": "2026-08-07",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "ibm-granite/granite-4.2-8b",
      "license": "Apache-2.0",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "granite-4.2-8b"
      ],
      "description": "IBM's Granite 4.2 8B: successor to 4.1 with reasoning controls added; 131K context, text-only. $0.06/$0.25 per Mtok on OpenRouter.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "ibm-granite/granite-4.2-8b",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "inception/mercury-2",
      "name": "Mercury 2",
      "family": "mercury",
      "lab": "inception",
      "release_date": "2026-03-04",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "mercury-2"
      ],
      "description": "First commercial reasoning diffusion LLM (dLLM): parallel generate-and-refine decoding, 1000+ tok/s on H100-class hardware, ~300ms reasoning latency. 128K context, up to 50K output. $0.25/$0.75 per Mtok, cache $0.025. reasoning_effort instant/low/medium/high.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "inception/mercury-2",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "inclusionai/ling-3-0-flash",
      "name": "Ling 3.0 Flash",
      "family": "ling",
      "lab": "inclusionai",
      "release_date": "2026-07-23",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "inclusionAI/Ling-3.0-flash",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "ling-3.0-flash"
      ],
      "description": "Ant Group's Ling 3.0-Flash: 124B total / 5.1B-active MoE, 262K context, open-sourced 2026-08-05. $0.021/$0.063 per Mtok on OpenRouter — among the cheapest frontier-adjacent listings.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "inclusionai/ling-3.0-flash",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "kwaipilot/kat-coder-pro-v2-5",
      "name": "KAT-Coder Pro v2.5",
      "family": "kat",
      "lab": "kwaipilot",
      "release_date": "2026-07-10",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "kat-coder-pro-v2.5"
      ],
      "description": "Kwai's flagship agentic coding model: 262K context, text-only, no thinking mode. $0.74/$2.96 per Mtok on OpenRouter.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "kwaipilot/kat-coder-pro-v2.5",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "liquid/lfm-2-5-2-6b",
      "name": "LFM2.5-2.6B",
      "family": "lfm",
      "lab": "liquid",
      "release_date": "2026-08-11",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "LiquidAI/LFM2.5-2.6B",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "lfm-2.5-2.6b"
      ],
      "description": "Liquid AI's compact 2.6B edge model that always reasons (thinking cannot be disabled); 64K context. Served free on OpenRouter; open weights (GGUF variants also published).",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "liquid/lfm-2.5-2.6b:free",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "meituan/longcat-2-0",
      "name": "LongCat 2.0",
      "family": "longcat",
      "lab": "meituan",
      "release_date": "2026-06-30",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "meituan-longcat/LongCat-2.0",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "LongCat-2.0"
      ],
      "description": "Meituan's 1.6T/48B-active sparse-MoE agentic coding model (MIT); 1M context; thinking toggleable via chat-template commands with thinking_budget.",
      "offered_via": [
        {
          "provider": "opencode-go",
          "wire_id": "longcat-2.0",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "longcat-2.0",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "meituan/longcat-2.0",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "meta/llama-3-3-70b",
      "name": "Llama 3.3 70B",
      "family": "llama",
      "lab": "meta",
      "release_date": "2024-12-06",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "Llama-3.3-70B-Instruct",
        "llama-v3p3-70b-instruct"
      ],
      "description": "Open-weight 70B instruct model; served by third parties.",
      "offered_via": [
        {
          "provider": "io-intelligence",
          "wire_id": "meta-llama/Llama-3.3-70B-Instruct",
          "endpoint": "chat"
        },
        {
          "provider": "nvidia",
          "wire_id": "meta/llama-3.3-70b-instruct",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "meta-llama/llama-3.3-70b-instruct",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "meta/llama-4-maverick",
      "name": "Llama 4 Maverick",
      "family": "llama",
      "lab": "meta",
      "release_date": "2025-04-27",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "Llama-4-Maverick-17B-128E-Instruct-FP8"
      ],
      "description": "Open-weight MoE (17B active/128 experts); the last Llama generation shipped before the Llama API retired 2026-07-06; served by third parties.",
      "ranking": {
        "aa_index": 14,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/llama-4-maverick",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "io-intelligence",
          "wire_id": "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",
          "endpoint": "chat"
        },
        {
          "provider": "openrouter",
          "wire_id": "meta-llama/llama-4-maverick",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "meta/muse-glimmer-30b",
      "name": "Muse Glimmer 30B",
      "family": "muse-glimmer",
      "lab": "meta",
      "release_date": "2026-08-10",
      "retired_date": "",
      "knowledge_cutoff": "2026-01-04",
      "open_weights": true,
      "hf_repo": "meta-models/Muse-Glimmer-30B",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "muse-glimmer-30b"
      ],
      "description": "Open-weight (Apache 2.0) 30B reasoning model, 128K context; served on NVIDIA NIM, not on Meta's own API. Knowledge cutoff per third-party model-card summaries.",
      "ranking": {
        "aa_index": 35,
        "aa_variant": "high",
        "url": "https://artificialanalysis.ai/models/muse-glimmer-30b",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "nvidia",
          "wire_id": "meta/muse-glimmer-30b",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "meta/muse-glimmer-30b",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "meta/muse-spark-1-2",
      "name": "Muse Spark 1.2",
      "family": "muse-spark",
      "lab": "meta",
      "release_date": "2026-08-05",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "audio",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "muse-spark-1.2"
      ],
      "description": "Meta's hosted flagship (served by the Meta Model API); always-on reasoning (reasoning_effort none returns 400); 1M context; image/video/PDF/audio input.",
      "ranking": {
        "aa_index": 57,
        "aa_variant": "xhigh effort",
        "url": "https://artificialanalysis.ai/models/muse-spark-1-2",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "meta",
          "wire_id": "muse-spark-1.2",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "meta",
          "wire_id": "muse-spark-1.2-contributor",
          "endpoint": "v1-responses"
        },
        {
          "provider": "meta",
          "wire_id": "muse-spark-1.2",
          "endpoint": "v1-responses"
        },
        {
          "provider": "opencode-go",
          "wire_id": "muse-spark-1.2-contributor",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "muse-spark-1.2",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "muse-spark-1.2",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "meta/muse-spark-1.2",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "meta/muse-spark-1-3",
      "name": "Muse Spark 1.3",
      "family": "muse-spark",
      "lab": "meta",
      "release_date": "2026-09-02",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "audio",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "muse-spark-1.3"
      ],
      "description": "Meta's flagship follow-up to Muse Spark 1.2, positioned as its frontier return: 1.05M context tuned for long-context workloads, and audio+video added to the input mix over 1.2's image/PDF. Always-on reasoning. $1.25/$4.25 per Mtok; a Contributor program SKU serves at $0.10/$0.20. Rolling out via Muse Code and the Meta Model API.",
      "offered_via": [
        {
          "provider": "opencode-go",
          "wire_id": "muse-spark-1.3-contributor",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "muse-spark-1.3-contributor-free",
          "endpoint": "chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "muse-spark-1.3",
          "endpoint": "responses"
        },
        {
          "provider": "openrouter",
          "wire_id": "meta/muse-spark-1.3-contributor",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "meta/muse-spark-1.3",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "microsoft/phi-4-mini-flash-reasoning",
      "name": "Phi-4 Mini Flash Reasoning",
      "family": "phi",
      "lab": "microsoft",
      "release_date": "",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "microsoft/Phi-4-mini-flash-reasoning",
      "license": "MIT",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "3.8B hybrid SSM+attention reasoning model, MIT; delisted from hosted catalogs as of 2026-08-20 — open weights only.",
      "offered_via": []
    },
    {
      "id": "minimax/minimax-m2-5",
      "name": "MiniMax M2.5",
      "family": "minimax-m",
      "lab": "minimax",
      "release_date": "2026-02-12",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "MiniMax-AI/MiniMax-M2.5",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "MiniMax-M2.5"
      ],
      "description": "Thinking cannot be turned off; thinking.type disabled has no effect.",
      "offered_via": [
        {
          "provider": "io-intelligence",
          "wire_id": "MiniMaxAI/MiniMax-M2.5",
          "endpoint": "chat"
        },
        {
          "provider": "minimax",
          "wire_id": "MiniMax-M2.5",
          "endpoint": "chat-completions"
        },
        {
          "provider": "minimax-token-plan",
          "wire_id": "MiniMax-M2.5",
          "endpoint": "anthropic"
        },
        {
          "provider": "opencode-go",
          "wire_id": "minimax-m2.5",
          "endpoint": "go-messages"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "minimax-m2.5",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "minimax/minimax-m2.5",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "minimax/minimax-m2-7",
      "name": "MiniMax M2.7",
      "family": "minimax-m",
      "lab": "minimax",
      "release_date": "",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "MiniMax-M2.7",
        "minimax-m2.7"
      ],
      "description": "229B open-weight model; replaced M2.5 on Ollama Cloud 2026-07-31. Release date not published.",
      "ranking": {
        "aa_index": 39,
        "aa_variant": "reasoning",
        "url": "https://artificialanalysis.ai/models/minimax-m2-7",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "io-intelligence",
          "wire_id": "MiniMaxAI/MiniMax-M2.7",
          "endpoint": "chat"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "minimax-m2.7",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "minimax-m2.7",
          "endpoint": "go-messages"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "minimax-m2.7",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "minimax/minimax-m2.7",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "minimax/minimax-m3",
      "name": "MiniMax M3",
      "family": "minimax-m",
      "lab": "minimax",
      "release_date": "2026-06-01",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "MiniMax-M3"
      ],
      "description": "Controllable adaptive thinking; disabled skips thinking.",
      "ranking": {
        "aa_index": 45,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/minimax-m3",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "minimax",
          "wire_id": "MiniMax-M3",
          "endpoint": "chat-completions"
        },
        {
          "provider": "minimax-token-plan",
          "wire_id": "MiniMax-M3",
          "endpoint": "anthropic"
        },
        {
          "provider": "nvidia",
          "wire_id": "minimaxai/minimax-m3",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "minimax-m3",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "minimax-m3",
          "endpoint": "go-messages"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "minimax-m3",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "minimax/minimax-m3",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "mistral/mistral-large-3",
      "name": "Mistral Large 3",
      "family": "mistral-large",
      "lab": "mistral",
      "release_date": "2025-12-02",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "",
      "license": "Apache-2.0",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "mistral-large-2512",
        "mistral-large-3-2512"
      ],
      "description": "675B/41B-active multimodal MoE flagship; $0.50/$1.50 per Mtok.",
      "ranking": {
        "aa_index": 16,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/mistral-large-3",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "ollama-cloud",
          "wire_id": "mistral-large-3:675b",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "mistralai/mistral-large-2512",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "mistral/mistral-medium-3-5",
      "name": "Mistral Medium 3.5",
      "family": "mistral-medium",
      "lab": "mistral",
      "release_date": "2026-04-28",
      "retired_date": "",
      "knowledge_cutoff": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "mistral-medium-3.5"
      ],
      "description": "Supports reasoning_effort high/none; Magistral reasoning models are deprecated.",
      "ranking": {
        "aa_index": 30,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/mistral-medium-3-5",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "mistral",
          "wire_id": "mistral-medium-3.5",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "mistralai/mistral-medium-3-5",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "mistral/mistral-small-latest",
      "name": "Mistral Small (latest)",
      "family": "mistral-small",
      "lab": "mistral",
      "release_date": "2026-03-16",
      "retired_date": "",
      "knowledge_cutoff": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Supports reasoning_effort high/none; with high, content becomes ThinkChunk + TextChunk blocks.",
      "offered_via": [
        {
          "provider": "mistral",
          "wire_id": "mistral-small-latest",
          "endpoint": "v1-chat-completions"
        }
      ]
    },
    {
      "id": "moonshot/kimi-k2-5",
      "name": "Kimi K2.5",
      "family": "kimi",
      "lab": "moonshot",
      "release_date": "2026-01-27",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "kimi-k2.5"
      ],
      "description": "Closed to new registrations; platform sunset 2026-08-31. Release date per third-party reporting.",
      "offered_via": [
        {
          "provider": "io-intelligence",
          "wire_id": "moonshotai/Kimi-K2.5",
          "endpoint": "chat"
        },
        {
          "provider": "moonshot",
          "wire_id": "kimi-k2.5",
          "endpoint": "chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "kimi-k2.5",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "kimi-k2.5",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "moonshotai/kimi-k2.5",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "moonshot/kimi-k2-6",
      "name": "Kimi K2.6",
      "family": "kimi",
      "lab": "moonshot",
      "release_date": "2026-04-20",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "kimi-k2.6"
      ],
      "description": "General-purpose multimodal with hybrid thinking (enabled/disabled) and Preserved Thinking (keep=all). Release date per third-party reporting.",
      "ranking": {
        "aa_index": 45,
        "aa_variant": "reasoning",
        "url": "https://artificialanalysis.ai/models/kimi-k2-6",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "fireworks-ai",
          "wire_id": "kimi-k2p6",
          "endpoint": "inference-v1"
        },
        {
          "provider": "io-intelligence",
          "wire_id": "moonshotai/Kimi-K2.6",
          "endpoint": "chat"
        },
        {
          "provider": "moonshot",
          "wire_id": "kimi-k2.6",
          "endpoint": "chat-completions"
        },
        {
          "provider": "near-ai",
          "wire_id": "moonshotai/kimi-k2.6",
          "endpoint": "chat"
        },
        {
          "provider": "nvidia",
          "wire_id": "moonshotai/kimi-k2.6",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "kimi-k2.6",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "kimi-k2.6",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "kimi-k2.6",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "moonshotai/kimi-k2.6",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "moonshot/kimi-k2-7-code",
      "name": "Kimi K2.7 Code",
      "family": "kimi",
      "lab": "moonshot",
      "release_date": "2026-06-12",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "kimi-k2.7-code"
      ],
      "description": "Current coding model (~180 tok/s); text/image/video input; thinking.type accepts only enabled. Release date per third-party reporting.",
      "ranking": {
        "aa_index": 43,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/kimi-k2-7-code",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "alibaba-token-plan",
          "wire_id": "kimi-k2.7-code",
          "endpoint": "compatible-mode"
        },
        {
          "provider": "io-intelligence",
          "wire_id": "moonshotai/Kimi-K2.7-Code",
          "endpoint": "chat"
        },
        {
          "provider": "kimi-coding",
          "wire_id": "kimi-for-coding-highspeed",
          "endpoint": "coding-openai"
        },
        {
          "provider": "kimi-coding",
          "wire_id": "kimi-for-coding",
          "endpoint": "coding-openai"
        },
        {
          "provider": "moonshot",
          "wire_id": "kimi-k2.7-code",
          "endpoint": "anthropic"
        },
        {
          "provider": "moonshot",
          "wire_id": "kimi-k2.7-code-highspeed",
          "endpoint": "chat-completions"
        },
        {
          "provider": "moonshot",
          "wire_id": "kimi-k2.7-code",
          "endpoint": "chat-completions"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "kimi-k2.7-code",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "kimi-k2.7-code",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "kimi-k2.7-code",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "moonshotai/kimi-k2.7-code",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "moonshot/kimi-k3",
      "name": "Kimi K3",
      "family": "kimi",
      "lab": "moonshot",
      "release_date": "2026-07-22",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "k3"
      ],
      "description": "Flagship 2.8T-param MoE (16/896 active) with vision; always-on reasoning; weights open-sourcing promised 2026-07-27.",
      "ranking": {
        "aa_index": 60,
        "aa_variant": "max effort",
        "url": "https://artificialanalysis.ai/models/kimi-k3",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "baseten",
          "wire_id": "moonshotai/Kimi-K3",
          "endpoint": "chat-completions"
        },
        {
          "provider": "io-intelligence",
          "wire_id": "moonshotai/Kimi-K3",
          "endpoint": "chat"
        },
        {
          "provider": "kimi-coding",
          "wire_id": "k3-256k",
          "endpoint": "coding-anthropic"
        },
        {
          "provider": "kimi-coding",
          "wire_id": "k3",
          "endpoint": "coding-openai"
        },
        {
          "provider": "moonshot",
          "wire_id": "kimi-k3",
          "endpoint": "chat-completions"
        },
        {
          "provider": "near-ai",
          "wire_id": "moonshotai/kimi-k3",
          "endpoint": "chat"
        },
        {
          "provider": "nvidia",
          "wire_id": "moonshotai/kimi-k3",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "kimi-k3",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "kimi-k3",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "kimi-k3",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "moonshotai/kimi-k3",
          "endpoint": "chat-completions"
        },
        {
          "provider": "synthetic",
          "wire_id": "hf:moonshotai/Kimi-K3",
          "endpoint": "openai-chat"
        },
        {
          "provider": "synthetic",
          "wire_id": "syn:large:vision",
          "endpoint": "openai-chat"
        }
      ]
    },
    {
      "id": "motif/motif-3",
      "name": "Motif-3",
      "family": "motif",
      "lab": "motif",
      "release_date": "2026-08-13",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "Motif-Technologies/Motif-3",
      "license": "MIT",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Korea sovereign-AI flagship (Motif Technologies): 314B/13.2B-active MoE, 256K context, MIT. First-party API serving unconfirmed — currently open-weights only.",
      "ranking": {
        "aa_index": 47,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/motif-3",
        "verified": "2026-08-20"
      },
      "offered_via": []
    },
    {
      "id": "nex-agi/nex-n2-5-mini",
      "name": "Nex N2.5 Mini",
      "family": "nex",
      "lab": "nex-agi",
      "release_date": "2026-09-08",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "nex-n2.5-mini"
      ],
      "description": "Compact tier of the N2.5 generation: 262K context, text+image input, reasoning-capable. Currently served only as OpenRouter's :free SKU; weights not yet public.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "nex-agi/nex-n2.5-mini:free",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "nex-agi/nex-n2-5-pro",
      "name": "Nex N2.5 Pro",
      "family": "nex",
      "lab": "nex-agi",
      "release_date": "2026-09-08",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "nex-n2.5-pro"
      ],
      "description": "Nex N2.5 Pro succeeds N2 Pro barely three months after it: 262K context, text+image input, reasoning-capable. Currently served only as OpenRouter's :free SKU; weights not yet public.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "nex-agi/nex-n2.5-pro:free",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "nex-agi/nex-n2-mini",
      "name": "Nex N2 Mini",
      "family": "nex",
      "lab": "nex-agi",
      "release_date": "2026-06-24",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "nex-agi/Nex-N2-mini",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "nex-n2-mini"
      ],
      "description": "Compact multimodal sibling of Nex N2 Pro; 262K context, text+image input, $0.025/$0.10 per Mtok on OpenRouter.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "nex-agi/nex-n2-mini",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "nex-agi/nex-n2-pro",
      "name": "Nex N2 Pro",
      "family": "nex",
      "lab": "nex-agi",
      "release_date": "2026-06-08",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "nex-agi/Nex-N2-Pro",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "nex-n2-pro"
      ],
      "description": "Agentic MoE flagship on the Qwen3.5 architecture: 397B total / 17B active, 262K context, text+image input. Open weights; $0.25/$1.00 per Mtok on OpenRouter.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "nex-agi/nex-n2-pro",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "nvidia/nemotron-3-5-lightning",
      "name": "Nemotron 3.5 Lightning",
      "family": "nemotron",
      "lab": "nvidia",
      "release_date": "2026-08-11",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "nemotron-3.5-lightning-30b-a3b"
      ],
      "description": "30B-A3B MoE + Mamba-2 hybrid attention, up to 1M context; NVFP4 checkpoint fits one GPU.",
      "ranking": {
        "aa_index": 24,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/nemotron-3-5-lightning",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "nvidia",
          "wire_id": "nvidia/nemotron-3.5-lightning-30b-a3b",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "nemotron-3.5-lightning-free",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "nvidia/nemotron-3.5-lightning",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "nvidia/nemotron-3-nano",
      "name": "Nemotron 3 Nano 30B A3B",
      "family": "nemotron",
      "lab": "nvidia",
      "release_date": "2025-12-15",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16",
      "license": "NVIDIA Nemotron Open Model License",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "nemotron-3-nano-30b-a3b"
      ],
      "description": "Fully open 30B/3.5B-active hybrid Mamba-2/MoE with released training data and recipes; 262K context (1M-capable config), up to 228K output. chat_template_kwargs.enable_thinking toggles thinking, default on (off mode uses greedy decoding). $0.05/$0.20 per Mtok.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "nvidia/nemotron-3-nano-30b-a3b",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "nvidia/nemotron-3-super",
      "name": "Nemotron 3 Super",
      "family": "nemotron",
      "lab": "nvidia",
      "release_date": "2026-03-11",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "nemotron-3-super-120b-a12b"
      ],
      "description": "120.6B/12.7B-active hybrid Mamba-Transformer MoE; up to 1M context (256K default); reasoning + tool calling.",
      "offered_via": [
        {
          "provider": "nvidia",
          "wire_id": "nvidia/nemotron-3-super-120b-a12b",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "nemotron-3-super",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "nvidia/nemotron-3-super-120b-a12b",
          "endpoint": "chat-completions"
        },
        {
          "provider": "synthetic",
          "wire_id": "hf:nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4",
          "endpoint": "openai-chat"
        }
      ]
    },
    {
      "id": "nvidia/nemotron-3-ultra",
      "name": "Nemotron 3 Ultra",
      "family": "nemotron",
      "lab": "nvidia",
      "release_date": "2026-06-04",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "nvidia/nemotron-3-ultra-550b-a55b",
        "NVIDIA-Nemotron-3-Ultra-550B-A55B"
      ],
      "description": "550B MoE (55B active) reasoning model; served on NVIDIA NIM and Baseten.",
      "ranking": {
        "aa_index": 38,
        "aa_variant": "550B-A55B",
        "url": "https://artificialanalysis.ai/models/nvidia-nemotron-3-ultra-550b-a55b",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "nvidia",
          "wire_id": "nvidia/nemotron-3-ultra-550b-a55b",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "nemotron-3-ultra",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "nemotron-3-ultra-free",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "nvidia/nemotron-3-ultra-550b-a55b",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "openai/gpt-5-1",
      "name": "GPT-5.1",
      "family": "gpt",
      "lab": "openai",
      "release_date": "2025-11-13",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Defaults to effort 'none'; supports none/low/medium/high.",
      "ranking": {
        "aa_index": 37,
        "aa_variant": "high effort",
        "url": "https://artificialanalysis.ai/models/gpt-5-1",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "openai/gpt-5.1",
          "endpoint": "chat"
        },
        {
          "provider": "openai",
          "wire_id": "gpt-5.1",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openai",
          "wire_id": "gpt-5.1",
          "endpoint": "v1-responses"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gpt-5.1",
          "endpoint": "responses"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5.1",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "openai/gpt-5-4-mini",
      "name": "GPT-5.4 mini",
      "family": "gpt",
      "lab": "openai",
      "release_date": "2026-03-17",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "gpt-5.4-mini"
      ],
      "description": "400K context, 128K output, text+image. $0.75/$4.50 per Mtok, cached input $0.075. reasoning.effort none-xhigh, default none.",
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "openai/gpt-5.4-mini",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gpt-5.4-mini",
          "endpoint": "responses"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5.4-mini",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "openai/gpt-5-4-nano",
      "name": "GPT-5.4 nano",
      "family": "gpt",
      "lab": "openai",
      "release_date": "2026-03-17",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "gpt-5.4-nano"
      ],
      "description": "400K context, 128K output, text+image. $0.20/$1.25 per Mtok, cached input $0.02. reasoning.effort none-xhigh, default none. Functionally succeeded by gpt-5.6-luna but served.",
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "openai/gpt-5.4-nano",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gpt-5.4-nano",
          "endpoint": "responses"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5.4-nano",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "openai/gpt-5-4",
      "name": "GPT-5.4",
      "family": "gpt",
      "lab": "openai",
      "release_date": "2026-03-05",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "gpt-5.4"
      ],
      "description": "1.05M context, 128K output, text+image input. $2.50/$15.00 per Mtok (prompts over 272K billed 2x input / 1.5x output for the session); cached input $0.25. reasoning.effort none-xhigh (no max), default none. Superseded positionally by 5.5/5.6 but served and not deprecated.",
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "openai/gpt-5.4",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gpt-5.4-pro",
          "endpoint": "responses"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gpt-5.4",
          "endpoint": "responses"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5.4-pro",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5.4",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "openai/gpt-5-5",
      "name": "GPT-5.5",
      "family": "gpt",
      "lab": "openai",
      "release_date": "2026-04-23",
      "retired_date": "",
      "knowledge_cutoff": "2025-12-01",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "gpt-5.5"
      ],
      "description": "Effort none–xhigh default medium; 1.05M context; prompts >272K billed 2x input/1.5x output. GPT-5.5-Pro variant $30/$180.",
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "openai/gpt-5.5",
          "endpoint": "chat"
        },
        {
          "provider": "openai",
          "wire_id": "gpt-5.5",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openai",
          "wire_id": "gpt-5.5",
          "endpoint": "v1-responses"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gpt-5.5-pro",
          "endpoint": "responses"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gpt-5.5",
          "endpoint": "responses"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5.5-pro",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5.5",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "openai/gpt-5-6",
      "name": "GPT-5.6",
      "family": "gpt",
      "lab": "openai",
      "release_date": "2026-06-25",
      "retired_date": "",
      "knowledge_cutoff": "2026-02-16",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Adds xhigh everywhere and max on Responses; Chat Completions rejects tools unless effort is none.",
      "ranking": {
        "aa_index": 61,
        "aa_variant": "Sol variant, max effort",
        "url": "https://artificialanalysis.ai/models/gpt-5-6-sol",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "azure-foundry",
          "wire_id": "gpt-5.6",
          "endpoint": "v1-responses"
        },
        {
          "provider": "near-ai",
          "wire_id": "openai/gpt-5.6-luna",
          "endpoint": "chat"
        },
        {
          "provider": "near-ai",
          "wire_id": "openai/gpt-5.6-sol",
          "endpoint": "chat"
        },
        {
          "provider": "openai",
          "wire_id": "gpt-5.6",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openai",
          "wire_id": "gpt-5.6-luna",
          "endpoint": "codex-backend"
        },
        {
          "provider": "openai",
          "wire_id": "gpt-5.6",
          "endpoint": "v1-responses"
        },
        {
          "provider": "openai",
          "wire_id": "gpt-5.6-sol",
          "endpoint": "codex-backend"
        },
        {
          "provider": "openai",
          "wire_id": "gpt-5.6-terra",
          "endpoint": "codex-backend"
        },
        {
          "provider": "opencode-go",
          "wire_id": "gpt-5.6-luna",
          "endpoint": "go-responses"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gpt-5.6-luna",
          "endpoint": "responses"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gpt-5.6-sol",
          "endpoint": "responses"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gpt-5.6-terra",
          "endpoint": "responses"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5.6-luna-pro",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5.6-luna",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5.6-sol-pro",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5.6-sol",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5.6-terra-pro",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5.6-terra",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "openai/gpt-5",
      "name": "GPT-5",
      "family": "gpt",
      "lab": "openai",
      "release_date": "2025-08-07",
      "retired_date": "",
      "knowledge_cutoff": "2024-09-30",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Reasoning model; default effort medium, no 'none'.",
      "ranking": {
        "aa_index": 35,
        "aa_variant": "high effort",
        "url": "https://artificialanalysis.ai/models/gpt-5",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "openai/gpt-5",
          "endpoint": "chat"
        },
        {
          "provider": "openai",
          "wire_id": "gpt-5",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openai",
          "wire_id": "gpt-5",
          "endpoint": "v1-responses"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gpt-5",
          "endpoint": "responses"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-5",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "openai/gpt-6",
      "name": "GPT-6 Astra",
      "family": "gpt",
      "lab": "openai",
      "release_date": "2026-09-03",
      "retired_date": "",
      "knowledge_cutoff": "2026-04-30",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "gpt-6-astra",
        "gpt-6"
      ],
      "description": "OpenAI's next-generation flagship, launched Sep 3 2026 (wire id gpt-6-astra; no plain gpt-6). 1.05M context (922K max input), 128K output, text+image input. $10/$50 per Mtok short-context ($20/$75 over 272K), cached input $1.00, cache writes $12.50; Batch/Flex $5/$25; fast mode $20/$100 (unavailable with EU data residency). Effort low-medium-high-xhigh-max with no none (thinking cannot be disabled); default undocumented. Chat Completions, Responses, and Batch only. Rolling out first to enterprises in the Trusted Access Program with API and plan access 'coming in the coming days' — not yet listed on any aggregator surface. OpenAI reports Astra as its first model to reach the 'Critical' cyber capability level; the strongest cyber features are restricted to vetted testers.",
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "openai/gpt-6-astra",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "gpt-6-astra",
          "endpoint": "responses"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-6-astra-pro",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-6-astra",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "openai/gpt-oss-120b",
      "name": "GPT-OSS 120B",
      "family": "gpt-oss",
      "lab": "openai",
      "release_date": "2025-08-05",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "openai/gpt-oss-120b",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "gpt-oss-120b"
      ],
      "description": "117B-active MoE open-weight reasoning model; thinking cannot be disabled (low/medium/high only).",
      "ranking": {
        "aa_index": 24,
        "aa_variant": "high effort",
        "url": "https://artificialanalysis.ai/models/gpt-oss-120b",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "baseten",
          "wire_id": "openai/gpt-oss-120b",
          "endpoint": "chat-completions"
        },
        {
          "provider": "fireworks-ai",
          "wire_id": "gpt-oss-120b",
          "endpoint": "inference-v1"
        },
        {
          "provider": "io-intelligence",
          "wire_id": "openai/gpt-oss-120b",
          "endpoint": "chat"
        },
        {
          "provider": "near-ai",
          "wire_id": "openai/gpt-oss-120b",
          "endpoint": "chat"
        },
        {
          "provider": "nvidia",
          "wire_id": "openai/gpt-oss-120b",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "gpt-oss:120b",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-oss-120b",
          "endpoint": "chat-completions"
        },
        {
          "provider": "synthetic",
          "wire_id": "hf:openai/gpt-oss-120b",
          "endpoint": "openai-chat"
        }
      ]
    },
    {
      "id": "openai/gpt-oss-20b",
      "name": "GPT-OSS 20B",
      "family": "gpt-oss",
      "lab": "openai",
      "release_date": "2025-08-05",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "openai/gpt-oss-20b",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "gpt-oss-20b"
      ],
      "description": "21B MoE open-weight reasoning model; thinking cannot be disabled.",
      "ranking": {
        "aa_index": 15,
        "aa_variant": "high effort",
        "url": "https://artificialanalysis.ai/models/gpt-oss-20b",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "io-intelligence",
          "wire_id": "openai/gpt-oss-20b",
          "endpoint": "chat"
        },
        {
          "provider": "nvidia",
          "wire_id": "openai/gpt-oss-20b",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "gpt-oss:20b",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/gpt-oss-20b",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "openai/o3",
      "name": "o3",
      "family": "o",
      "lab": "openai",
      "release_date": "2025-04-16",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "o-series reasoning model; effort low/medium/high.",
      "ranking": {
        "aa_index": 31,
        "aa_variant": "estimated",
        "url": "https://artificialanalysis.ai/models/o3",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "near-ai",
          "wire_id": "openai/o3",
          "endpoint": "chat"
        },
        {
          "provider": "openai",
          "wire_id": "o3",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openai",
          "wire_id": "o3",
          "endpoint": "v1-responses"
        },
        {
          "provider": "openrouter",
          "wire_id": "openai/o3",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "poolside/laguna-s-2-1",
      "name": "Laguna S 2.1",
      "family": "laguna",
      "lab": "poolside",
      "release_date": "",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "poolside/Laguna-S-2.1",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "118B/8B MoE, 1M context, 30T tokens; agentic coding.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "poolside/laguna-s-2.1",
          "endpoint": "chat-completions"
        },
        {
          "provider": "poolside",
          "wire_id": "poolside/laguna-s-2.1",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "poolside/laguna-xs-2-1",
      "name": "Laguna XS 2.1",
      "family": "laguna",
      "lab": "poolside",
      "release_date": "",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "poolside/Laguna-XS-2.1",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "33B/3B MoE, 256K context.",
      "offered_via": [
        {
          "provider": "nvidia",
          "wire_id": "poolside/laguna-xs-2.1",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "poolside/laguna-xs-2.1",
          "endpoint": "chat-completions"
        },
        {
          "provider": "poolside",
          "wire_id": "poolside/laguna-xs-2.1",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "sakana/fugu-ultra",
      "name": "Fugu Ultra",
      "family": "fugu",
      "lab": "sakana",
      "release_date": "2026-06-24",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "fugu-ultra"
      ],
      "description": "Sakana AI's multi-agent-as-a-model entry: 1M context, text+image input, reasoning_effort control. $5/$30 per Mtok on OpenRouter. Weights gated/not public.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "sakana/fugu-ultra",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "stepfun/step-3-5-flash",
      "name": "Step-3.5 Flash",
      "family": "step",
      "lab": "stepfun",
      "release_date": "2026-01-29",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "stepfun-ai/Step-3.5-Flash",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "196B/11B MoE, 256K ctx (3:1 sliding:full attention), MTP-3 decoding, Parallel Thinking reasoning.",
      "ranking": {
        "aa_index": 27,
        "aa_variant": "estimated",
        "url": "https://artificialanalysis.ai/models/step-3-5-flash",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "stepfun/step-3.5-flash",
          "endpoint": "chat-completions"
        },
        {
          "provider": "stepfun",
          "wire_id": "step-3.5-flash",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "stepfun/step-3-7-flash",
      "name": "Step-3.7 Flash",
      "family": "step",
      "lab": "stepfun",
      "release_date": "",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "stepfun-ai/Step-3.7-Flash",
      "license": "Apache-2.0",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "step-3.7-flash"
      ],
      "description": "198B sparse-MoE VLM, 256K ctx, Apache 2.0; three reasoning levels; Advisor Mode.",
      "ranking": {
        "aa_index": 31,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/step-3-7-flash",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "nvidia",
          "wire_id": "stepfun-ai/step-3.7-flash",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "stepfun/step-3.7-flash",
          "endpoint": "chat-completions"
        },
        {
          "provider": "stepfun",
          "wire_id": "step-3.7-flash",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "tencent/hy3",
      "name": "Hy3",
      "family": "hy3",
      "lab": "tencent",
      "release_date": "2026-07-06",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "tencent/Hy3",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Tencent Hunyuan 295B/21B-active MoE, 256K context, 3.8B MTP layer; default no-think plus low/high CoT modes; agentic/coding focus. GA 2026-07-06; preview shipped 2026-04-24.",
      "ranking": {
        "aa_index": 42,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/hy3",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "opencode-go",
          "wire_id": "hy3-preview",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "hy3",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "tencent/hy3",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "tencent/hy4",
      "name": "Hy4 Preview",
      "family": "hy4",
      "lab": "tencent",
      "release_date": "2026-08-28",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "tencent/Hy4-preview",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "hy4-preview",
        "hy4"
      ],
      "description": "Tencent's next-gen Hunyuan flagship, open-sourced as a preview on 2026-08-28: 770B total / 49B-active MoE with a 1M context window, FP8 variant shipped alongside. Successor to Hy3 (295B/21B).",
      "offered_via": [
        {
          "provider": "opencode-go",
          "wire_id": "hy4-preview",
          "endpoint": "go-chat-completions"
        }
      ]
    },
    {
      "id": "thinkingmachines/inkling-small",
      "name": "Inkling Small",
      "family": "inkling",
      "lab": "thinkingmachines",
      "release_date": "2026-07-29",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "audio"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "thinkingmachines/inkling-small"
      ],
      "description": "Smaller Inkling variant.",
      "ranking": {
        "aa_index": 41,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/inkling-small",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "baseten",
          "wire_id": "thinkingmachines/inkling-small",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "thinkingmachines/inkling-small",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "thinkingmachines/inkling",
      "name": "Inkling",
      "family": "inkling",
      "lab": "thinkingmachines",
      "release_date": "2026-07-15",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "audio"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "thinkingmachines/inkling"
      ],
      "description": "Thinking Machines' reasoning model with vision and audio input, 1M context / 32K output; served on Baseten.",
      "ranking": {
        "aa_index": 42,
        "aa_variant": "base",
        "url": "https://artificialanalysis.ai/models/inkling",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "baseten",
          "wire_id": "thinkingmachines/inkling",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "thinkingmachines/inkling",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "upstage/solar-pro-4",
      "name": "Solar Pro 4",
      "family": "solar",
      "lab": "upstage",
      "release_date": "2026-08-10",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "solar-pro4",
        "solar-pro4-20260810"
      ],
      "description": "Upstage's current flagship reasoning model (Korean-optimized); 524K context, 131K output, tool calling. List $0.30/$1.20 per Mtok with a 90%-off promo ($0.03/$0.12) through ~2026-09-10; cache read $0.006. Reasoning supported; native control parameter undocumented. Params undisclosed; open sibling is Solar-Open2-250B.",
      "offered_via": [
        {
          "provider": "openrouter",
          "wire_id": "upstage/solar-pro4",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "xai/grok-4-5",
      "name": "Grok 4.5",
      "family": "grok",
      "lab": "xai",
      "release_date": "2026-07-16",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "grok-4.5"
      ],
      "description": "Reasoning model; effort low/medium/high default high, cannot disable. xhigh silently treated as high.",
      "ranking": {
        "aa_index": 56,
        "aa_variant": "high effort",
        "url": "https://artificialanalysis.ai/models/grok-4-5",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "opencode-go",
          "wire_id": "grok-4.5",
          "endpoint": "go-responses"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "grok-4.5",
          "endpoint": "responses"
        },
        {
          "provider": "openrouter",
          "wire_id": "x-ai/grok-4.5",
          "endpoint": "chat-completions"
        },
        {
          "provider": "xai",
          "wire_id": "grok-4.5",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "xai",
          "wire_id": "grok-4.5",
          "endpoint": "v1-responses"
        }
      ]
    },
    {
      "id": "xai/grok-4-6",
      "name": "Grok 4.6",
      "family": "grok",
      "lab": "xai",
      "release_date": "2026-08-12",
      "retired_date": "",
      "knowledge_cutoff": "2026-02-01",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "grok-4.6"
      ],
      "description": "Reasoning model; adds xhigh over 4.5; returns summarized reasoning via streamed reasoning_content deltas.",
      "ranking": {
        "aa_index": 61,
        "aa_variant": "high effort",
        "url": "https://artificialanalysis.ai/models/grok-4-6",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "opencode-go",
          "wire_id": "grok-4.6",
          "endpoint": "go-responses"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "grok-4.6",
          "endpoint": "responses"
        },
        {
          "provider": "openrouter",
          "wire_id": "x-ai/grok-4.6",
          "endpoint": "chat-completions"
        },
        {
          "provider": "xai",
          "wire_id": "grok-4.6",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "xai",
          "wire_id": "grok-4.6",
          "endpoint": "v1-responses"
        }
      ]
    },
    {
      "id": "xiaomi/mimo-v2-5-pro",
      "name": "MiMo V2.5 Pro",
      "family": "mimo",
      "lab": "xiaomi",
      "release_date": "2026-04-23",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "XiaomiMiMo/MiMo-V2.5-Pro",
      "license": "MIT",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "1.02T/42B-active MoE agentic flagship, 1M context; GDPVal-AA #1 open-source; ultraspeed FP4 variant >1000 tok/s early access.",
      "ranking": {
        "aa_index": 43,
        "aa_variant": "reasoning",
        "url": "https://artificialanalysis.ai/models/mimo-v2-5-pro",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "opencode-go",
          "wire_id": "mimo-v2.5-pro",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "xiaomi/mimo-v2.5-pro",
          "endpoint": "chat-completions"
        },
        {
          "provider": "xiaomi-mimo",
          "wire_id": "mimo-v2.5-pro",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "xiaomi/mimo-v2-5",
      "name": "MiMo V2.5",
      "family": "mimo",
      "lab": "xiaomi",
      "release_date": "2026-04-23",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "XiaomiMiMo/MiMo-V2.5",
      "license": "MIT",
      "modalities": {
        "input": [
          "text",
          "image",
          "video",
          "audio"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Xiaomi's omni-modal open-weight line (text/image/video/audio), 1M context, MIT; V2 series retired 2026-06-30. Weights open-sourced 2026-06-29.",
      "ranking": {
        "aa_index": 38,
        "aa_variant": "reasoning, 0424 build",
        "url": "https://artificialanalysis.ai/models/mimo-v2-5-0424",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "io-intelligence",
          "wire_id": "XiaomiMiMo/MiMo-V2.5",
          "endpoint": "chat"
        },
        {
          "provider": "opencode-go",
          "wire_id": "mimo-v2.5",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "xiaomi/mimo-v2.5",
          "endpoint": "chat-completions"
        },
        {
          "provider": "xiaomi-mimo",
          "wire_id": "mimo-v2.5",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "zai/glm-4-6",
      "name": "GLM-4.6",
      "family": "glm",
      "lab": "zai",
      "release_date": "2025-09-30",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "zai-org/GLM-4.6",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Hybrid thinking, on by default, disableable via thinking.type.",
      "ranking": {
        "aa_index": 29,
        "aa_variant": "Reasoning",
        "url": "https://artificialanalysis.ai/models/glm-4-6-reasoning",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "io-intelligence",
          "wire_id": "zai-org/GLM-4.6",
          "endpoint": "chat"
        },
        {
          "provider": "openrouter",
          "wire_id": "z-ai/glm-4.6",
          "endpoint": "chat-completions"
        },
        {
          "provider": "zai",
          "wire_id": "glm-4.6",
          "endpoint": "chat-completions"
        },
        {
          "provider": "zai-coding-plan",
          "wire_id": "glm-4.6",
          "endpoint": "coding-chat-completions"
        }
      ]
    },
    {
      "id": "zai/glm-4-7-flash",
      "name": "GLM-4.7 Flash",
      "family": "glm",
      "lab": "zai",
      "release_date": "2026-01-19",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "zai-org/GLM-4.7-Flash",
      "license": "MIT",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "glm-4.7-flash"
      ],
      "description": "Small fast open-weight sibling of GLM-4.7 (MIT weights); ~200K context, text-only. $0.06/$0.40 per Mtok on OpenRouter; serves as Synthetic's syn:small:text default.",
      "offered_via": [
        {
          "provider": "io-intelligence",
          "wire_id": "zai-org/GLM-4.7-Flash",
          "endpoint": "chat"
        },
        {
          "provider": "openrouter",
          "wire_id": "z-ai/glm-4.7-flash",
          "endpoint": "chat-completions"
        },
        {
          "provider": "synthetic",
          "wire_id": "hf:zai-org/GLM-4.7-Flash",
          "endpoint": "openai-chat"
        },
        {
          "provider": "synthetic",
          "wire_id": "syn:small:text",
          "endpoint": "openai-chat"
        }
      ]
    },
    {
      "id": "zai/glm-4-7",
      "name": "GLM-4.7",
      "family": "glm",
      "lab": "zai",
      "release_date": "2025-12-22",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "glm-4.7",
        "glm-4p7",
        "GLM-4.7"
      ],
      "description": "Open-weight GLM generation between 4.6 and 5.x; hybrid thinking via enable_thinking on self-host-style surfaces.",
      "ranking": {
        "aa_index": 34,
        "aa_variant": "Reasoning",
        "url": "https://artificialanalysis.ai/models/glm-4-7",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "fireworks-ai",
          "wire_id": "glm-4p7",
          "endpoint": "inference-v1"
        },
        {
          "provider": "io-intelligence",
          "wire_id": "zai-org/GLM-4.7",
          "endpoint": "chat"
        },
        {
          "provider": "openrouter",
          "wire_id": "z-ai/glm-4.7",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "zai/glm-5-1",
      "name": "GLM-5.1",
      "family": "glm",
      "lab": "zai",
      "release_date": "2026-04-07",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "zai-org/GLM-5.1",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "glm-5.1"
      ],
      "description": "744B/40B-active MoE (MIT); thinking default enabled; 200K context; superseded by 5.2/5.3 but still listed and priced.",
      "offered_via": [
        {
          "provider": "io-intelligence",
          "wire_id": "zai-org/GLM-5.1",
          "endpoint": "chat"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "glm-5.1",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "glm-5.1",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "glm-5.1",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "z-ai/glm-5.1",
          "endpoint": "chat-completions"
        },
        {
          "provider": "zai",
          "wire_id": "glm-5.1",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "zai/glm-5-2",
      "name": "GLM-5.2",
      "family": "glm",
      "lab": "zai",
      "release_date": "2026-06-16",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "glm-5.2"
      ],
      "description": "756B open-weight model, 976K context.",
      "ranking": {
        "aa_index": 53,
        "aa_variant": "max",
        "url": "https://artificialanalysis.ai/models/glm-5-2",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "alibaba-token-plan",
          "wire_id": "glm-5.2",
          "endpoint": "compatible-mode"
        },
        {
          "provider": "baseten",
          "wire_id": "zai-org/GLM-5.2",
          "endpoint": "chat-completions"
        },
        {
          "provider": "io-intelligence",
          "wire_id": "zai-org/GLM-5.2",
          "endpoint": "chat"
        },
        {
          "provider": "near-ai",
          "wire_id": "z-ai/glm-5.2",
          "endpoint": "chat"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "glm-5.2",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "glm-5.2",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "glm-5.2",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "z-ai/glm-5.2",
          "endpoint": "chat-completions"
        },
        {
          "provider": "synthetic",
          "wire_id": "hf:zai-org/GLM-5.2",
          "endpoint": "openai-chat"
        },
        {
          "provider": "synthetic",
          "wire_id": "syn:large:text",
          "endpoint": "openai-chat"
        }
      ]
    },
    {
      "id": "zai/glm-5-3-flash",
      "name": "GLM-5.3 Flash",
      "family": "glm",
      "lab": "zai",
      "release_date": "2026-08-26",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": true,
      "hf_repo": "zai-org/GLM-5.3-Flash",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image",
          "video"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [
        "glm-5.3-flash"
      ],
      "description": "First natively multimodal GLM-5-series (image/video/file in); 320B/18B-active sparse+linear attention; 1M context; thinking cannot be disabled.",
      "offered_via": [
        {
          "provider": "io-intelligence",
          "wire_id": "zai-org/GLM-5.3-Flash",
          "endpoint": "chat"
        },
        {
          "provider": "near-ai",
          "wire_id": "z-ai/glm-5.3-flash",
          "endpoint": "chat"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "glm-5.3-flash",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "glm-5.3-flash",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "glm-5.3-flash",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "z-ai/glm-5.3-flash",
          "endpoint": "chat-completions"
        },
        {
          "provider": "zai",
          "wire_id": "glm-5.3-flash",
          "endpoint": "chat-completions"
        }
      ]
    },
    {
      "id": "zai/glm-5-3",
      "name": "GLM-5.3",
      "family": "glm",
      "lab": "zai",
      "release_date": "2026-08-18",
      "retired_date": "",
      "knowledge_cutoff": "",
      "open_weights": false,
      "hf_repo": "",
      "license": "",
      "modalities": {
        "input": [
          "text",
          "image"
        ],
        "output": [
          "text"
        ]
      },
      "aliases": [],
      "description": "Flagship GLM; forced thinking — thinking.type enabled/disabled both error.",
      "ranking": {
        "aa_index": 60,
        "aa_variant": "max effort",
        "url": "https://artificialanalysis.ai/models/glm-5-3",
        "verified": "2026-08-20"
      },
      "offered_via": [
        {
          "provider": "io-intelligence",
          "wire_id": "zai-org/GLM-5.3",
          "endpoint": "chat"
        },
        {
          "provider": "ollama-cloud",
          "wire_id": "glm-5.3",
          "endpoint": "v1-chat-completions"
        },
        {
          "provider": "opencode-go",
          "wire_id": "glm-5.3",
          "endpoint": "go-chat-completions"
        },
        {
          "provider": "opencode-zen",
          "wire_id": "glm-5.3",
          "endpoint": "chat-completions"
        },
        {
          "provider": "openrouter",
          "wire_id": "z-ai/glm-5.3",
          "endpoint": "chat-completions"
        },
        {
          "provider": "zai",
          "wire_id": "glm-5.3",
          "endpoint": "chat-completions"
        },
        {
          "provider": "zai-coding-plan",
          "wire_id": "glm-5.3",
          "endpoint": "coding-chat-completions"
        }
      ]
    }
  ]
}