GOT-OCR 2.0 vs Typhoon OCR
Pick GOT-OCR 2.0 for self-hosted work on formulas, sheet music, chemical structures or box-guided region OCR in English and Chinese. Pick Typhoon OCR for Thai or Thai-English documents such as government forms, financial reports and books, where it has published scores and an API option. The main split is language and maturity: GOT-OCR 2.0 lists no benchmark scores and has open issues with looping and wrong text on blank pages, while Typhoon OCR is built for Thai and English only.
GOT-OCR 2.0stepfun/got-ocr-2 | Typhoon OCRtyphoon/typhoon-ocr | |
|---|---|---|
| Type | Open source | Open source |
| Price | 1 credit / page | 2 credits / page |
| Per 1,000 pages | $1.00 | $2.00 |
| Typical speed | 1.5 s / page | 1.8 s / page |
| Best for | Formatted text, formulas and charts | Thai documents |
| Inputs | images | PDF, images |
| Languages | Chinese, English | Thai, English |
| Max pages per request | 100 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | No |
| Math | ✓ Yes | No |
| Bounding boxes | ✓ Yes | 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 GOT-OCR 2.0 if
- It outputs LaTeX, TikZ, SMILES and kern, so it handles formulas, chemical structures and sheet music that Typhoon OCR does not list.
- It supports region OCR guided by box coordinates or color, so you can read one part of a page.
- It runs in Hugging Face Transformers with batched inference, dynamic resolution and multi-page OCR, at 580M parameters.
Choose Typhoon OCR if
- It is built for Thai and Thai-English documents, and V1.5 scored BLEU 0.870 on Thai government forms, ahead of GPT-5 and Gemini 2.5 Pro.
- It gives you a choice of self-hosting the weights or calling the Typhoon API as typhoon-ocr-v2.
- It returns tables as HTML with merged cells, equations as LaTeX and marked checkboxes, and V2 2B scores 84.7 on olmOCR-bench.
Key differences
| GOT-OCR 2.0 | Typhoon OCR | |
|---|---|---|
| Languages | English and Chinese documents | Thai and English only; no other language evaluated |
| How you run it | Self-host only; CUDA 11.8 and PyTorch 2.0.1 per README, with a llama.cpp port for CPU | Self-host the weights or use the Typhoon API (typhoon-ocr-v2) |
| Licence | Code Apache-2.0; training data CC BY-NC 4.0 | Apache-2.0, subject to OpenTyphoon Terms and Conditions |
| Output formats | Plain text, Markdown, LaTeX, TikZ, SMILES, kern | Markdown, HTML, LaTeX |
| Model size | 580M parameters | 2B (V2 and V1.5); 0.8B V2 variant |
| Published benchmarks | Not published | olmOCR-bench 84.7, OmniDocBench v1.6 90.97, ThaiOCRBench 59.9 (V2 2B, by the authors) |
| Known failures | Loops characters midway; returns wrong text on images with no text; format mode is much worse than plain OCR mode | V2 can loop until the token limit on dense multi-column pages; key-information extraction scores only 21.0 on ThaiOCRBench |
Sources
- GOT-OCR2.0 GitHub
- GOT-OCR 2.0 paper
- GitHub issue index
- Typhoon OCR V2 2B model card
- Typhoon OCR technical report
What each one is
GOT-OCR 2.0
Unified OCR model that reads plain text, tables, formulas, sheet music and charts.
GOT-OCR 2.0 detailsTyphoon 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": "stepfun/got-ocr-2","fallbacks": ["typhoon/typhoon-ocr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
GOT-OCR 2.0 alternativesTyphoon OCR alternativesBest OCR models in 2026