Nougat vs Florence-2
Pick Nougat only for English scientific papers where you need Markdown with LaTeX math and tables, and only if your use is non-commercial. Pick Florence-2 for short text in images, scene text or edge devices where you need text with boxes under an MIT licence. The main split is output: Nougat converts a page to structured Markdown, while Florence-2 returns plain text and bounding boxes.
Nougatmeta/nougat | Florence-2microsoft/florence-2 | |
|---|---|---|
| Type | Open source | Open source |
| Price | 2 credits / page | 1 credit / page |
| Per 1,000 pages | $2.00 | $1.00 |
| Typical speed | 3.0 s / page | 0.8 s / page |
| Best for | arXiv-style papers | Lightweight OCR with boxes |
| Inputs | PDF, images | images |
| Languages | Mainly English | English |
| Max pages per request | 200 | 500 |
| Tables | ✓ Yes | No |
| Handwriting | No | No |
| Math | ✓ Yes | No |
| Bounding boxes | No | ✓ Yes |
| Confidence | No | No |
| Forms | No | No |
| Multilingual | No | No |
| On OCRRouter | Available | Available |
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
- Choose Nougat when you need LaTeX equations and tables in Mathpix-style Markdown from arXiv-style papers.
- Choose Nougat for math-heavy English PDFs, since it was trained on arXiv and PubMed Central documents.
- Choose Nougat if you want a simple CLI and API with small and base checkpoints for a non-commercial project.
Choose Florence-2 if
- Choose Florence-2 if you need a commercial-friendly MIT licence for both code and model.
- Choose Florence-2 when you need quadrilateral boxes per text line from OCR_WITH_REGION.
- Choose Florence-2 for lightweight English OCR on edge devices, at 0.23B (base) or 0.77B (large) parameters, or when the same model must also do captioning, detection and grounding.
Key differences
| Nougat | Florence-2 | |
|---|---|---|
| Licence | Code MIT; weights CC-BY-NC, so no commercial use | MIT |
| Output formats | Markdown and LaTeX (Mathpix-compatible .mmd) | Plain text and bounding boxes; no Markdown, tables or LaTeX |
| Best document type | English arXiv-style scientific papers and math-heavy PDFs | Scene text and short labels; not for full-page document conversion |
| Model size | Small and base checkpoints; parameter count not published | 0.23B (base), 0.77B (large) |
| Languages | Works best on English; Chinese, Russian and Japanese do not work | Chinese output mostly wrong because the vocabulary lacks CJK characters; spelling errors on French text in one test |
| Known failure | Repetition loops on 1.5% of test pages in the paper, more on out-of-domain documents; repetition detection can drop pages as MISSING_PAGE | Spelling errors and missing spaces on cursive handwriting; not suited to handwriting |
| Price and hosting | Free for non-commercial use, self-hosted (you pay for compute) | Free, self-hosted via Transformers (you pay for compute) |
Sources
- Nougat GitHub README
- Nougat paper
- Issue #11 repetitions
- Florence-2-large model card
- HF discussion Chinese very bad
What each one is
Florence-2
Small Microsoft vision model (0.23B and 0.77B) with OCR and OCR-with-regions tasks that return text with boxes.
Florence-2 detailsTry both on your own file
Run the file once with each model and compare the output.
PDF, PNG, JPG, WebP, HEIC. Up to 10 files at onceNo file?
Use one, fall back to the other
{"model": "meta/nougat","fallbacks": ["microsoft/florence-2"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Nougat vs Tesseract.jsNougat vs Typhoon OCRDonut vs Florence-2MonkeyOCR vs Florence-2GPT vs Florence-2Florence-2 vs OCR.spaceLlamaParse vs Florence-2Nougat vs OpenOCR
Nougat alternativesFlorence-2 alternativesBest OCR models in 2026