Vision

Image analysis and OCR: ask a question about an image, get a text answer.

Request

http

POST /v1/vision/analyze

json

{
  "prompt": "Que voit-on sur cette image ? Reponds en une phrase.",
  "image_url": "https://exemple.com/photo.jpg",
  "model": null,
  "routing_strategy": "auto"
}

image_url must be a publicly accessible URL. To analyze a file you just uploaded, first go through the upload flow (see POST /v1/uploads/presign) then use the returned URL.

Response

json

{
  "text": "Un chat roux assis sur le rebord d'une fenetre, regardant dehors.",
  "model": "gemini-3.1-pro-preview",
  "provider": "google"
}

Common use cases

Image description for accessibility
Text extraction (OCR) — explicitly ask to transcribe the visible text
Analysis of scanned documents, invoices, forms

Providers

OpenAI (GPT-4o and vision models), Google (Gemini).