OCRRouter

Nougat vs Pixtral

Pick Nougat for English scientific papers where you need LaTeX math and tables, and only for non-commercial work. Pick Pixtral for question answering over charts and documents, mainly if you already use the Mistral API. Pixtral is deprecated, so do not start a new OCR project on it.

Nougatmeta/nougatPixtralmistral/pixtral
TypeOpen sourceVision model
Price2 credits / page2 credits / page
Per 1,000 pages$2.00$2.00
Typical speed3.0 s / page3.0 s / page
Best forarXiv-style papersQuestions about document images
InputsPDF, imagesimages
LanguagesMainly EnglishMultilingual
Max pages per request200500
Tables✓ YesNo
HandwritingNoNo
Math✓ YesNo
Bounding boxesNoNo
ConfidenceNoNo
FormsNoNo
MultilingualNo✓ Yes
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 Nougat if

  • It is trained on arXiv and PubMed Central papers and outputs LaTeX for equations and tables in Mathpix-style Markdown.
  • It is a small self-hosted Python package with small and base checkpoints, a simple CLI and an API.
  • Code is MIT and it is free to run for research, so you only pay for compute.

Choose Pixtral if

  • Apache-2.0 open weights allow commercial use, and you can also call it through the Mistral API at a launch price of $0.15 per million input and output tokens.
  • It scores 90.7 ANLS on DocVQA and 81.8 on ChartQA, according to Mistral, which suits question answering over documents and charts.
  • Its 128K token context allows many images in one request, and it can return plain text, Markdown or JSON.

Key differences

NougatPixtral
LicenceCode MIT; weights CC-BY-NC, so no commercial useApache-2.0
How you run itSelf-hosted Python package onlyMistral API (pixtral-12b-2409) or self-hosted open weights
PriceFree for non-commercial use; you pay for compute$0.15 per million input and output tokens at API launch; free to self-host
Output formatsMarkdown (Mathpix-style .mmd) and LaTeXPlain text, Markdown and JSON
BenchmarksNot publishedDocVQA 90.7 ANLS, ChartQA 81.8, MMMU 52.5 (reported by Mistral)
Languages and handwritingWorks best on English; Chinese, Russian and Japanese do not workReported accuracy drop on handwriting, small fonts and non-Latin scripts
Known failureRepetition loops on 1.5% of test pages, more on out-of-domain documents; looping pages can be dropped as MISSING_PAGEDoes not always get OCR near correct; Q4 quantization degrades small text; Mistral marks it deprecated and no longer maintained

Sources

What each one is

Nougat

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

Nougat details

Pixtral

Mistral's open-weight vision models, Pixtral 12B and Pixtral Large. Mistral's API no longer serves them, so they run self-hosted.

Pixtral 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": "meta/nougat","fallbacks": ["mistral/pixtral"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

Nougat alternativesPixtral alternativesBest OCR models in 2026