فا

Document OCR

POST /v1/ocr — a document or image in, markdown pages out; priced per page.

POSThttps://iranrouter.com/v1/ocr

Document OCR: a PDF or image (a URL or a data URI) in, markdown per page out — Mistral OCR's shape. Priced per page processed; the rate is on the model's page.

Shell
curl https://iranrouter.com/v1/ocr \
  -H "Authorization: Bearer ir-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "mistralai/mistral-ocr-latest",
    "document": { "type": "document_url", "document_url": "https://example.com/contract.pdf" },
    "pages": [0, 1, 2]
  }'
JSON
{
  "pages": [ { "index": 0, "markdown": "# ...", "images": [], "dimensions": { "dpi": 200, "height": 2200, "width": 1700 } } ],
  "model": "mistralai/mistral-ocr-latest",
  "usage_info": { "pages_processed": 3, "doc_size_bytes": 184230 },
  "usage": { "cost_rial": 45000 }
}
ParameterNotes
document{type: document_url, document_url} or {type: image_url, image_url}; a data URI works too.
pagesPage numbers (from zero). Without it the whole document is read.
include_image_base64Return the extracted images as base64.

The wallet hold before processing is the pages you named in pages, else 10 pages; the final charge is computed from usage_info.pages_processed.