Baseten
first_party
Getting credentials
api-key
api_key via header (header Authorization: Bearer)
Environment: BASETEN_API_KEY
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.
Endpoints
| ID | URL | Protocol | Auth |
|---|---|---|---|
| chat-completions | https://inference.baseten.co/v1/chat/completions | openai-chat | any |
| messages | https://inference.baseten.co/v1/messages | anthropic-messages | any |
API surfaces
textstreaming
Quirks
- Baseten reads Authorization only — the Anthropic SDK's default x-api-key is NOT read; override default_headers. (docs)
- reasoning_effort is validated (invalid values 400); chat_template_args.enable_thinking on opt-in models is unvalidated and fails silently. (docs)
- Rate limits are account-wide (Basic verified 120 RPM / 500k TPM); 529 = overloaded; GET /v1/models embeds pricing and context. (docs)
- The Anthropic-compatible /v1/messages endpoint is in beta; behavior may change before general availability — prefer Chat Completions for production workloads. (docs)
Models offered
| Model | Wire ID | Endpoint | Status | Reasoning |
|---|---|---|---|---|
| deepseek/deepseek-v4-flash | deepseek-ai/DeepSeek-V4-Flash-0731 | chat-completions | ga | effort · default on · returns reasoning_content |
| moonshot/kimi-k3 | moonshotai/Kimi-K3 | chat-completions | ga | effort · default on · returns reasoning_content · round-trips reasoning_content |
| openai/gpt-oss-120b | openai/gpt-oss-120b | chat-completions | ga | effort · default on · returns reasoning_content |
| thinkingmachines/inkling | thinkingmachines/inkling | chat-completions | ga | effort · default on · returns reasoning_content · round-trips reasoning_content |
| thinkingmachines/inkling-small | thinkingmachines/inkling-small | chat-completions | ga | effort · default on · returns reasoning_content · round-trips reasoning_content |
| zai/glm-5-2 | zai-org/GLM-5.2 | chat-completions | ga | effort · default on · returns reasoning_content |