Text recognition
Convert scanned pages to text. Check the page rate and whether layout metadata is included.
OCR and document extraction
Plain text OCR, layout parsing, annotated extraction and token-priced vision LLMs are not interchangeable. Define the required output, then estimate the charge for that task.
Select the output
Convert scanned pages to text. Check the page rate and whether layout metadata is included.
Extract sections, tables, or form fields. Managed document processors may price different features separately.
Request bounding boxes or richer extraction. Check model capability and its applicable annotated-page rate.
Image-capable language models can instead bill for image inputs and generated tokens. Do not treat a per-page OCR rate as a token price or assume equal extraction quality. If you send PDFs directly to OpenAI, Claude or Gemini, use the PDF input cost guide to estimate their different billing units.
Official model examples
API ID mistral-ocr-4-1. Standard: $0.004/page; annotated: $0.005/annotated page.
API ID mistral-ocr-4-0. Standard: $0.004/page; annotated: $0.005/annotated page.
Managed document-processing route. Region and pricing tier may differ; a catalog rate alone is not a like-for-like quote.
The mistral-ocr-latest alias is not an independent provider offer. Do not infer its future version or rank these routes by price without matching capabilities.
Worked budget
10,000 × $0.004 = $40
10,000 × $0.005 = $50
Billable page count × verified per-page rate. Check any additional services separately.
Illustrative list-price totals, not an invoice. These exclude discounts, storage, preparation, taxes, retries and unverified cache or Batch eligibility. Do not add the annotated and standard rates without confirming the provider's billing rules.
For adjacent services, Google Document AI lists different processor pricing for OCR, layout and extraction. It is not an interchangeable LLM route. Browse OCR catalog entries for source-backed model details. The token calculator is not a native page-priced OCR estimator.