Nougat vs Typhoon OCR
Pick Nougat for English arXiv-style papers with heavy math, when your use is non-commercial and you can self-host. Pick Typhoon OCR for Thai or Thai-English documents such as government forms and financial reports, or when you need an API and commercial-friendly terms. The main split is domain and licence: Nougat is tied to English science papers under CC-BY-NC weights, while Typhoon OCR targets Thai business documents under Apache-2.0 plus the OpenTyphoon terms.
Nougatmeta/nougat | Typhoon OCRtyphoon/typhoon-ocr | |
|---|---|---|
| Type | Open source | Open source |
| Price | 2 credits / page | 2 credits / page |
| Per 1,000 pages | $2.00 | $2.00 |
| Typical speed | 3.0 s / page | 1.8 s / page |
| Best for | arXiv-style papers | Thai documents |
| Inputs | PDF, images | PDF, images |
| Languages | Mainly English | Thai, English |
| Max pages per request | 200 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | No |
| Math | ✓ Yes | No |
| Bounding boxes | No | No |
| Confidence | No | No |
| Forms | No | No |
| Multilingual | No | ✓ Yes |
| 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
- Your input is English arXiv-style scientific papers, which is what it was trained on (arXiv and PubMed Central).
- You need math-heavy PDFs converted to LaTeX, with equations and tables in Mathpix-compatible .mmd Markdown.
- Your use is non-commercial and you want a small or base checkpoint with a simple CLI and API.
Choose Typhoon OCR if
- Your documents are Thai or Thai-English mixed: government forms, financial reports or books.
- You want tables as HTML with merged cells, equations as LaTeX and checkboxes marked, in Markdown, HTML or LaTeX output.
- You want to either self-host the 2B model (0.8B V2 variant also exists) or call it on the Typhoon API as typhoon-ocr-v2.
Key differences
| Nougat | Typhoon OCR | |
|---|---|---|
| Licence | Code MIT; weights CC-BY-NC, so no commercial use | Apache-2.0, subject to OpenTyphoon Terms and Conditions |
| How you run it | Self-hosted Python package only | Self-hosted weights, or the Typhoon API as typhoon-ocr-v2 |
| Languages | Works best on English; README says Chinese, Russian and Japanese do not work | Thai and English only were evaluated |
| Output formats | Markdown and LaTeX | Markdown, HTML and LaTeX |
| Benchmarks | Not published | V2 2B: olmOCR-bench 84.7, OmniDocBench v1.6 90.97, ThaiOCRBench 59.9 (authors' figures) |
| Known failure: loops | Repetition loops on 1.5% of test pages, more on out-of-domain documents; repetition detection can drop pages as MISSING_PAGE | V2 can loop until the token limit, mostly on dense multi-column pages |
| Weakest area | Complex layouts, where attention loses position and causes loops; not suited to business documents, forms and invoices | Key-information extraction, scored 21.0 on ThaiOCRBench; can output text not in the image |
Sources
- Nougat GitHub README
- Nougat paper
- Issue #11 repetitions
- Typhoon OCR V2 2B model card
- Typhoon OCR technical report
What each one is
Typhoon OCR
Open vision model tuned for Thai and English documents, with Markdown output.
Typhoon OCR detailsTry both on your own file
Run the file once with each model and compare the output.
Use one, fall back to the other
{"model": "meta/nougat","fallbacks": ["typhoon/typhoon-ocr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Nougat alternativesTyphoon OCR alternativesBest OCR models in 2026