Typhoon OCR vs Pixtral
Pick Typhoon OCR for Thai and Thai-English documents such as government forms, financial reports and books, where it is built for the job and scores 84.7 on olmOCR-bench at 2B parameters. Pick Pixtral only if you already run it on the Mistral API and need question answering over charts and documents. Mistral marks Pixtral 12B as deprecated, so it is a poor base for a new OCR project.
Typhoon OCRtyphoon/typhoon-ocr | Pixtralmistral/pixtral | |
|---|---|---|
| Type | Open source | Vision model |
| Price | 2 credits / page | 2 credits / page |
| Per 1,000 pages | $2.00 | $2.00 |
| Typical speed | 1.8 s / page | 3.0 s / page |
| Best for | Thai documents | Questions about document images |
| Inputs | PDF, images | images |
| Languages | Thai, English | Multilingual |
| Max pages per request | 500 | 500 |
| Tables | ✓ Yes | No |
| Handwriting | No | No |
| Math | No | No |
| Bounding boxes | No | No |
| Confidence | No | No |
| Forms | No | No |
| Multilingual | ✓ Yes | ✓ 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 Typhoon OCR if
- You work with Thai or Thai-English documents: V1.5 beat GPT-5 and Gemini 2.5 Pro on Thai government forms (BLEU 0.870).
- You need structured output: tables as HTML with merged cells, equations as LaTeX, and checkboxes marked.
- You want a small model to host: the main V2 model is 2B parameters, with a 0.8B V2 variant.
Choose Pixtral if
- You already have an integration on the Mistral API and only need to keep it running.
- You want to ask questions about charts and documents: it scores 90.7 ANLS on DocVQA and 81.8 on ChartQA, per Mistral.
- You need many images in one request, since the context window is 128K tokens.
Key differences
| Typhoon OCR | Pixtral | |
|---|---|---|
| Licence | Apache-2.0, subject to OpenTyphoon Terms and Conditions | Apache-2.0 |
| Model size | 2B (V2 and V1.5); 0.8B V2 variant | 12.4B (12B decoder plus 400M vision encoder) |
| How you run it | Self-hosted weights, or the Typhoon API as typhoon-ocr-v2 | Open weights, or the Mistral API as pixtral-12b-2409 |
| Price | Free to self-host. Typhoon API pricing: Not published | $0.15 per million input and output tokens at Mistral API launch; free to self-host |
| Output formats | Markdown, HTML, LaTeX | Plain text, Markdown, JSON |
| Languages | Thai and English only evaluated | Reported accuracy drop on non-Latin scripts |
| Known failure | V2 can loop until the token limit on dense multi-column pages; weakest at key-information extraction (21.0 on ThaiOCRBench) | Deprecated by Mistral; a launch thread user reported it does not always get OCR close to correct; Q4 quantization degrades small text |
Sources
- Typhoon OCR V2 2B model card
- Typhoon OCR technical report
- Mistral announcement
- Mistral Sept 2024 pricing
- HN launch thread
What each one is
Typhoon OCR
Open vision model tuned for Thai and English documents, with Markdown output.
Typhoon OCR 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": "typhoon/typhoon-ocr","fallbacks": ["mistral/pixtral"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Typhoon OCR alternativesPixtral alternativesBest OCR models in 2026