Mistral
first_party
Getting credentials
api-key
api_key via header (header Authorization: Bearer)
Environment: MISTRAL_API_KEY
Create a key at console.mistral.ai (La Plateforme).
Endpoints
| ID | URL | Protocol | Auth |
|---|---|---|---|
| v1-chat-completions | https://api.mistral.ai/v1/chat/completions | openai-chat | any |
API surfaces
textstreamingembeddingsfilesbatch
Quirks
- 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)
Models offered
| Model | Wire ID | Endpoint | Status | Reasoning |
|---|---|---|---|---|
| mistral/mistral-medium-3-5 | mistral-medium-3.5 | v1-chat-completions | ga | effort · default off · returns thinking_blocks |
| mistral/mistral-small-latest | mistral-small-latest | v1-chat-completions | ga | effort · default off · returns thinking_blocks |