mistral_vision
Multimodal chat with images (URL or base64).
How to use it
mistral_vision is exposed by the Mistral MCP MCP server. Add the server to your MCP client (Claude Desktop, Cursor, Windsurf and others), and the mistral_vision tool becomes available to the model automatically. See the full listing for setup details and every tool this server provides.
Install Mistral MCP
claude mcp add mistral -- npx -y mistral-mcp@latestOther tools in Mistral MCP (12)
Fill-in-the-middle code completion (Codestral model).
Invoke a connector's tool — real MCP CallToolResult passthrough.
Fetch one connector's public metadata (never credentials).
Discover Mistral Connectors (MCP/HTTP integrations) visible to the caller.
List the MCP tools a connector exposes, with their input schema.
Chat completion. Supports all Mistral models, responseformat, reasoningeffort for Magistral.
Document AI — extract text, bbox, and JSON annotations from PDFs/images. Pass includeBlocks: true for OCR 4 paragraph-level blocks (text/title/table/image/equation/... with bounding boxes).
Single-call macro-tool: OCR → classify (kind=auto) → typed extraction → validation → cache. Kinds: contract / invoice / iddocument / generic. Returns a discriminated union. PII-safe cache (iddocument auto-bypass). Configurable minOcrConfidence.
Audio → text. Pass diarize: true for speaker separation.
Start a Mistral Workflow (Temporal-backed durable execution).
Signal / query a running workflow. Used for human-in-the-loop checkpoints.
Poll a running workflow — returns RUNNING \