{
  "source_commit": "2dc3a78",
  "generated_at": "2026-09-11T00:56:04.735Z",
  "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."
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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)."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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)."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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"
      }
    },
    {
      "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."
    },
    {
      "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"
      }
    }
  ]
}