openapi: 3.1.0
info:
  title: Mistral AI Agents OCR API
  description: Agent completions API for building AI agents that can handle complex tasks, maintain context, coordinate multiple actions, and use tools including function calling. Agents are created and configured via the Mistral platform and invoked through this API.
  version: '1.0'
  contact:
    name: Mistral AI Support
    url: https://docs.mistral.ai/
    email: support@mistral.ai
  termsOfService: https://mistral.ai/terms/
servers:
- url: https://api.mistral.ai/v1
  description: Mistral AI Production
security:
- bearerAuth: []
tags:
- name: OCR
  description: Document OCR and text extraction operations
paths:
  /ocr:
    post:
      operationId: processDocument
      summary: Mistral AI Process a document with OCR
      description: Extract text, images, tables, and structured data from documents, PDFs, or images. Returns structured Markdown with extracted content including mathematical expressions in LaTeX format.
      tags:
      - OCR
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/OcrRequest'
      responses:
        '200':
          description: OCR extraction result
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/OcrResponse'
        '400':
          description: Bad request
        '401':
          description: Unauthorized
        '429':
          description: Rate limit exceeded
components:
  schemas:
    ExtractedImage:
      type: object
      properties:
        id:
          type: string
          description: Image identifier referenced in the Markdown
        top_left_x:
          type: integer
        top_left_y:
          type: integer
        bottom_right_x:
          type: integer
        bottom_right_y:
          type: integer
        image_base64:
          type: string
          description: Base64-encoded image data if requested
    OcrRequest:
      type: object
      required:
      - model
      - document
      properties:
        model:
          type: string
          description: The OCR model to use
          examples:
          - mistral-ocr-latest
        document:
          $ref: '#/components/schemas/DocumentInput'
        pages:
          type: array
          items:
            type: integer
          description: Specific page numbers to process (0-indexed)
        include_image_base64:
          type: boolean
          default: false
          description: Whether to include extracted images as base64
        image_limit:
          type: integer
          description: Maximum number of images to extract
        image_min_size:
          type: integer
          description: Minimum image dimension in pixels to extract
    OcrPage:
      type: object
      properties:
        index:
          type: integer
          description: Zero-based page index
        markdown:
          type: string
          description: Extracted content in Markdown format
        images:
          type: array
          items:
            $ref: '#/components/schemas/ExtractedImage'
        dimensions:
          type: object
          properties:
            width:
              type: integer
            height:
              type: integer
    DocumentInput:
      type: object
      required:
      - type
      properties:
        type:
          type: string
          enum:
          - document_url
          - image_url
          - file_id
          description: The type of document input
        document_url:
          type: string
          format: uri
          description: URL of the document to process
        image_url:
          type: string
          format: uri
          description: URL of the image to process
        file_id:
          type: string
          description: ID of a previously uploaded file
    OcrResponse:
      type: object
      properties:
        pages:
          type: array
          items:
            $ref: '#/components/schemas/OcrPage'
        model:
          type: string
          description: The model used for OCR
        usage:
          type: object
          properties:
            pages_processed:
              type: integer
            doc_size_bytes:
              type: integer
  securitySchemes:
    bearerAuth:
      type: http
      scheme: bearer
      description: Mistral AI API key passed as a Bearer token
externalDocs:
  description: Mistral AI Agents API Documentation
  url: https://docs.mistral.ai/api/endpoint/agents