Mistral OCR vs Granite Docling
Pick Mistral OCR for high-volume PDF to Markdown through a hosted API, especially scanned business documents processed in batch. Pick Granite Docling for English PDFs with tables, code and equations that you want to run yourself under Apache-2.0 on a CPU or small GPU. The main split is a $2 per 1,000 pages hosted API versus a free 258M model you host and guard against repetition loops.
Mistral OCRmistral/mistral-ocr | Granite Doclingibm/granite-docling | |
|---|---|---|
| Type | OCR API | Open source |
| Price | 25 credits / page | 1 credit / page |
| Per 1,000 pages | $25.00 | $1.00 |
| Typical speed | 0.9 s / page | 1.5 s / page |
| Best for | High-volume PDF to Markdown | Small-model document conversion |
| Inputs | PDF, images | PDF, images |
| Languages | 100+ languages | English, experimental CJK and Arabic |
| Max pages per request | 1,000 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | No |
| Math | ✓ Yes | ✓ Yes |
| Bounding boxes | ✓ Yes | ✓ Yes |
| Confidence | No | No |
| Forms | No | No |
| Multilingual | ✓ Yes | 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 Mistral OCR if
- It is a hosted API at $2 per 1,000 pages, or $1 per 1,000 through the Batch API, so there is nothing to deploy.
- OCR 3 rebuilds tables as HTML with rowspan and colspan, and returns Markdown, HTML, JSON and extracted images.
- It has no published language limits in the sources, while Granite Docling marks Japanese, Arabic and Chinese as experimental.
Choose Granite Docling if
- It is Apache-2.0 and free to self-host, so you pay only for compute and your documents stay on your own machines.
- At 258M parameters it runs on a CPU or small GPU, inside the Docling library.
- It scores 0.97 TEDS table structure on FinTabNet, 0.968 equation F1 and 0.988 code F1 on the model card, so it suits technical English PDFs.
Key differences
| Mistral OCR | Granite Docling | |
|---|---|---|
| How you run it | Hosted API (mistral-ocr-2512, OCR 3) | Self-hosted, typically inside Docling |
| Licence | Proprietary | Apache-2.0 |
| Price | $2 per 1,000 pages; $3 with structured annotations; $1 via Batch API | Free; you pay for compute |
| Output formats | Markdown, HTML, JSON, images | DocTags, Markdown, HTML, LaTeX, bounding boxes |
| Benchmark | OmniDocBench v1.5 overall 78.83 for the first version; v1.6 overall 85.66 (independent) | Full-page OCR F1 0.84, OCRBench 500 (published by the authors) |
| Languages | Not published | English primary; Japanese, Arabic and Chinese experimental |
| Known failure | Hallucinated text on unreadable handwriting; changed checkbox states and misread characters on a dense financial table; colored-background pages returned as images in the 2025 release | Repetitive loops on page headings and invoices; exclamation-mark output on GPUs without bfloat16 such as T4 unless run in float32 |
Sources
- Mistral OCR 3 announcement
- VentureBeat on OCR 3 pricing
- OmniDocBench v1.6 leaderboard
- Reducto test of Mistral OCR
- Granite-Docling model card
- HF discussion #24 repetition
What each one is
Mistral OCR
Fast document OCR that returns clean Markdown with tables and embedded images.
Mistral OCR detailsGranite Docling
Compact IBM vision model that converts pages to DocTags for the Docling pipeline.
Granite Docling 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": "mistral/mistral-ocr","fallbacks": ["ibm/granite-docling"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Mistral OCR alternativesGranite Docling alternativesBest OCR models in 2026