GOT-OCR 2.0 vs Pixtral
Pick GOT-OCR 2.0 for self-hosted work on formulas, sheet music, chemical structures or region-specific OCR, mainly because it emits LaTeX, TikZ, SMILES and kern. Pick Pixtral only if you already call it through the Mistral API for question answering over charts and documents. Mistral marks Pixtral 12B as deprecated, so avoid it for new OCR projects.
GOT-OCR 2.0stepfun/got-ocr-2 | Pixtralmistral/pixtral | |
|---|---|---|
| Type | Open source | Vision model |
| Price | 1 credit / page | 2 credits / page |
| Per 1,000 pages | $1.00 | $2.00 |
| Typical speed | 1.5 s / page | 3.0 s / page |
| Best for | Formatted text, formulas and charts | Questions about document images |
| Inputs | images | images |
| Languages | Chinese, English | Multilingual |
| Max pages per request | 100 | 500 |
| Tables | ✓ Yes | No |
| 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 formulas, chemical structures and sheet music come out in a usable format.
- Region OCR takes box coordinates or color, so you can read one part of a page.
- At 580M parameters it is far smaller than Pixtral's 12.4B, and it is merged into Hugging Face Transformers with batched inference.
Choose Pixtral if
- It has a hosted Mistral API at a launch price of $0.15 per million input and output tokens, so you need no GPU of your own.
- Mistral reports 90.7 ANLS on DocVQA and 81.8 on ChartQA, which suits question answering over documents and charts.
- Its 128K token context allows many images in one request, and the weights are Apache-2.0.
Key differences
| GOT-OCR 2.0 | Pixtral | |
|---|---|---|
| Deployment | Self-hosted only | Mistral API (pixtral-12b-2409) or self-hosted open weights |
| Licence | Code Apache-2.0; training data CC BY-NC 4.0 | Apache-2.0 |
| Size | 580M parameters | 12.4B (12B decoder + 400M vision encoder) |
| Price | Free, self-hosted (you pay for compute) | $0.15 per million input and output tokens at API launch; free to self-host |
| Output formats | Plain text, Markdown, LaTeX, TikZ, SMILES, kern | Plain text, Markdown, JSON |
| Benchmarks | Not published | DocVQA 90.7 ANLS, ChartQA 81.8, MMMU 52.5 (reported by Mistral) |
| Known failure | Can loop characters midway through recognition, returns wrong text on blank images, and format mode scores much worse than plain OCR mode | Does not always get OCR near correct, and accuracy drops on handwriting, small fonts and non-Latin scripts; Mistral marks it deprecated |
Sources
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 detailsPixtral
Mistral's open-weight vision models, Pixtral 12B and Pixtral Large. Mistral's API no longer serves them, so they run self-hosted.
Pixtral 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": ["mistral/pixtral"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
GOT-OCR 2.0 alternativesPixtral alternativesBest OCR models in 2026