OCRRouter

Mistral OCR vs Nougat

Pick Mistral OCR for high-volume PDF to Markdown, scanned business documents and any commercial product, because it is a hosted API at $2 per 1,000 pages ($1 via Batch API) with a commercial-friendly service. Pick Nougat only for English arXiv-style papers where you need LaTeX math and can run it non-commercially on your own GPU. Nougat's weights are CC-BY-NC and it is not built for forms, invoices or non-Latin scripts.

Mistral OCRmistral/mistral-ocrNougatmeta/nougat
TypeOCR APIOpen source
Price25 credits / page2 credits / page
Per 1,000 pages$25.00$2.00
Typical speed0.9 s / page3.0 s / page
Best forHigh-volume PDF to MarkdownarXiv-style papers
InputsPDF, imagesPDF, images
Languages100+ languagesMainly English
Max pages per request1,000200
Tables✓ Yes✓ Yes
HandwritingNoNo
Math✓ Yes✓ Yes
Bounding boxes✓ YesNo
ConfidenceNoNo
FormsNoNo
Multilingual✓ YesNo
On OCRRouterAvailableAvailable

Prices are what you pay on OCRRouter. Speed is the provider's typical figure, not a benchmark we ran.

Which to choose

Choose Mistral OCR if

  • You want a hosted API with no GPU to run, priced at $2 per 1,000 pages, or $1 per 1,000 through the Batch API.
  • You process scanned business documents, forms, tables and handwriting in bulk, and OCR 3 rebuilds tables as HTML with rowspan and colspan.
  • You are building a commercial product, since Nougat's weights are CC-BY-NC and bar commercial use.

Choose Nougat if

  • Your input is English scientific papers like arXiv or PubMed Central documents, which are what it was trained on.
  • You need equations and tables as LaTeX in Mathpix-compatible .mmd Markdown, for math-heavy PDFs.
  • Your use is non-commercial and you want to self-host with a small or base checkpoint, a simple CLI and API, and pay only for compute.

Key differences

Mistral OCRNougat
How you run itHosted API (current pinned version mistral-ocr-2512, OCR 3)Self-hosted Python package
LicenceProprietaryCode MIT; weights CC-BY-NC, so no commercial use
Price$2 per 1,000 pages; $3 with structured annotations; $1 per 1,000 via Batch APIFree for non-commercial use; you pay for compute
Output formatsMarkdown, HTML, JSON, imagesMarkdown, LaTeX
Best document typesScanned business documents and high-volume PDF to Markdown for RAGEnglish arXiv-style scientific papers and math-heavy PDFs
BenchmarkOmniDocBench v1.5 overall 78.83 (Mistral OCR 2503); v1.6 overall 85.66Not published
Known failureHallucinated text on unreadable handwriting; colored-background pages returned as images with no OCR text in the 2025 releaseRepetition loops on 1.5% of test pages, more on out-of-domain documents; looped pages can be dropped as MISSING_PAGE
LanguagesNot publishedWorks best on English; Chinese, Russian and Japanese do not work

Sources

What each one is

Mistral OCR

Fast document OCR that returns clean Markdown with tables and embedded images.

Mistral OCR details

Nougat

Meta's model that converts academic PDFs to Markdown with LaTeX math.

Nougat details

Try both on your own file

Run the file once with each model and compare the output.

PDF, PNG, JPG, WebP, HEIC. Up to 20 MB each, 10 files.No file?

Use one, fall back to the other

{"model": "mistral/mistral-ocr","fallbacks": ["meta/nougat"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

Mistral OCR alternativesNougat alternativesBest OCR models in 2026