OCRRouter

Granite Docling vs Nougat

Pick Granite Docling for English PDFs with tables, code and equations, especially if you ship a commercial product: it is Apache-2.0 and only 258M parameters. Pick Nougat only for non-commercial work on English arXiv-style papers where you want LaTeX math in Mathpix-style Markdown, because its weights are CC-BY-NC.

Granite Doclingibm/granite-doclingNougatmeta/nougat
TypeOpen sourceOpen source
Price1 credit / page2 credits / page
Per 1,000 pages$1.00$2.00
Typical speed1.5 s / page3.0 s / page
Best forSmall-model document conversionarXiv-style papers
InputsPDF, imagesPDF, images
LanguagesEnglish, experimental CJK and ArabicMainly English
Max pages per request500200
Tables✓ Yes✓ Yes
HandwritingNoNo
Math✓ Yes✓ Yes
Bounding boxes✓ YesNo
ConfidenceNoNo
FormsNoNo
MultilingualNoNo
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 Granite Docling if

  • The Apache-2.0 licence allows commercial use, while Nougat's weights are CC-BY-NC.
  • It outputs DocTags, Markdown, HTML, LaTeX and bounding boxes, so it fits Docling pipelines and layout-aware work.
  • At 258M parameters it suits low-cost CPU or small GPU conversion, with a table structure TEDS of 0.97 on FinTabNet.

Choose Nougat if

  • It is trained on arXiv and PubMed Central documents, so it is built for academic papers.
  • It writes LaTeX for equations and tables in Mathpix-compatible .mmd Markdown, which suits math-heavy PDFs.
  • It ships small and base checkpoints with a simple CLI and API, and it is free for non-commercial use.

Key differences

Granite DoclingNougat
LicenceApache-2.0Code MIT; weights CC-BY-NC (no commercial use)
How you run itSelf-hosted, typically inside the Docling librarySelf-hosted Python package with CLI and API
Output formatsDocTags, Markdown, HTML, LaTeX, bounding boxesMarkdown (.mmd) and LaTeX
Best document typeEnglish PDFs with tables, code and equationsEnglish arXiv-style scientific papers
Published benchmarksTable TEDS 0.97 on FinTabNet, full-page OCR F1 0.84, equation F1 0.968Not published
LanguagesEnglish primary; Japanese, Arabic and Chinese are experimentalWorks best on English; README says Chinese, Russian and Japanese do not work
Known failureRepetitive loops on page headings and invoices; may output only exclamation marks on GPUs without bfloat16 such as T4 unless run in float32Repetition loops on 1.5% of test pages, more on out-of-domain documents; looping pages can be dropped as MISSING_PAGE

Sources

What each one is

Granite Docling

Compact IBM vision model that converts pages to DocTags for the Docling pipeline.

Granite Docling 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": "ibm/granite-docling","fallbacks": ["meta/nougat"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

Granite Docling alternativesNougat alternativesBest OCR models in 2026