Mistral OCR vs Florence-2
Pick Mistral OCR for high-volume PDF to Markdown, such as RAG ingestion or batches of scanned business documents. Pick Florence-2 only if you need a small, free, self-hosted model for short English text with bounding boxes, or one that also does detection. Florence-2 has no document structure output, so it cannot replace Mistral OCR for full pages.
Mistral OCRmistral/mistral-ocr | Florence-2microsoft/florence-2 | |
|---|---|---|
| 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 | 0.8 s / page |
| Best for | High-volume PDF to Markdown | Lightweight OCR with boxes |
| Inputs | PDF, images | images |
| Languages | 100+ languages | English |
| Max pages per request | 1,000 | 500 |
| Tables | ✓ Yes | No |
| Handwriting | No | No |
| Math | ✓ Yes | No |
| 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
- You need Markdown with tables from full pages; OCR 3 rebuilds tables as HTML with rowspan and colspan.
- You process large volumes and want a low price: $2 per 1,000 pages, or $1 per 1,000 through the Batch API.
- You want a hosted API with no GPU to run, with Markdown, HTML, JSON and extracted images as output.
Choose Florence-2 if
- You must self-host under the MIT licence and pay only for compute.
- You need per-line quadrilateral boxes from OCR_WITH_REGION, or one model for OCR, captioning, detection and grounding.
- You run on constrained hardware, since the model is 0.23B (base) or 0.77B (large) parameters, for English scene text and short labels.
Key differences
| Mistral OCR | Florence-2 | |
|---|---|---|
| How you run it | Hosted API (current pinned version mistral-ocr-2512, OCR 3) | Self-hosted via Transformers |
| Licence | Proprietary | MIT |
| Price | $2 per 1,000 pages; $3 with structured annotations; $1 per 1,000 via Batch API | Free; you pay for compute |
| Output formats | Markdown, HTML, JSON, images | Plain text and bounding boxes only |
| Tables and document structure | Tables rebuilt as HTML with rowspan and colspan | No Markdown, tables or LaTeX |
| Model size | Not published | 0.23B (base), 0.77B (large) |
| Known failure | Hallucinated text on unreadable handwriting; changed checkbox states and misread characters on a dense financial table page | Spelling errors and missing spaces on cursive handwriting and French text; Chinese output mostly wrong because the vocabulary lacks CJK characters |
Sources
- Mistral OCR 3 announcement
- VentureBeat on OCR 3 pricing
- Reducto test of Mistral OCR
- Florence-2-large model card
- HF discussion Chinese very bad
What each one is
Mistral OCR
Fast document OCR that returns clean Markdown with tables and embedded images.
Mistral OCR detailsFlorence-2
Small Microsoft vision model (0.23B and 0.77B) with OCR and OCR-with-regions tasks that return text with boxes.
Florence-2 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": ["microsoft/florence-2"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Mistral OCR alternativesFlorence-2 alternativesBest OCR models in 2026