OCRRouter

Qianfan OCR vs Nougat

Pick Qianfan OCR for commercial products, multilingual documents and mixed layouts such as tables, handwriting and scene text. Pick Nougat only for non-commercial work on English scientific papers where you want LaTeX math in Mathpix-style Markdown. The main split is licence and scope: Qianfan OCR is Apache-2.0 and covers 192 languages, while Nougat's weights are CC-BY-NC and it works best on English.

Qianfan OCRbaidu/qianfan-ocrNougatmeta/nougat
TypeOCR APIOpen source
Price2 credits / page2 credits / page
Per 1,000 pages$2.00$2.00
Typical speed2.0 s / page3.0 s / page
Best forChinese documentsarXiv-style papers
InputsPDF, imagesPDF, images
LanguagesChinese, English and moreMainly English
Max pages per request200200
Tables✓ Yes✓ Yes
Handwriting✓ YesNo
Math✓ Yes✓ Yes
Bounding boxesNoNo
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 Qianfan OCR if

  • The Apache-2.0 licence allows commercial use, so you can ship it in a product.
  • It covers 192 languages, handwriting and scene text, while Nougat's README says Chinese, Russian and Japanese do not work.
  • It outputs Markdown, JSON, HTML, LaTeX and bounding boxes, and suits key information extraction, tables and formulas.

Choose Nougat if

  • It is built for English arXiv-style papers, trained on arXiv and PubMed Central documents.
  • It turns math-heavy PDFs into LaTeX equations and tables inside Mathpix-compatible .mmd Markdown.
  • It is a small self-hosted Python package with small and base checkpoints and a simple CLI and API, free for non-commercial use.

Key differences

Qianfan OCRNougat
LicenceApache-2.0, commercial use allowedCode MIT; weights CC-BY-NC, no commercial use
How you run itSelf-host the open weights, or use the Baidu Qianfan platformSelf-host only, as a Python package
Model size4B parameters (Qwen3-4B language model)Not published
Output formatsMarkdown, JSON, HTML, LaTeX, bounding boxesMarkdown (.mmd) and LaTeX
Languages192 languages, plus handwriting and scene textWorks best on English; Chinese, Russian and Japanese do not work
Benchmarks93.12 on OmniDocBench v1.5, 79.8 on olmOCR-Bench, 880 on OCRBench (author-reported)Not published
Known failureThinking mode can emit up to 16,384 tokens, adding latency; 1.024 pages per second on an A100 even with W8A8Repetition 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

Qianfan OCR

Baidu's OCR-tuned vision model. Strong on Chinese documents and tables.

Qianfan 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": "baidu/qianfan-ocr","fallbacks": ["meta/nougat"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

Qianfan OCR alternativesNougat alternativesBest OCR models in 2026