GOT-OCR 2.0 vs Florence-2
Pick GOT-OCR 2.0 for formulas, tables, sheet music, chemical structures and English or Chinese documents where you need Markdown or LaTeX out. Pick Florence-2 for English scene text and short labels where you need line boxes, or where one small model must also do captioning and detection. The main split is structured document output versus a tiny, MIT-licensed model that returns only text and boxes.
GOT-OCR 2.0stepfun/got-ocr-2 | Florence-2microsoft/florence-2 | |
|---|---|---|
| Type | Open source | Open source |
| Price | 1 credit / page | 1 credit / page |
| Per 1,000 pages | $1.00 | $1.00 |
| Typical speed | 1.5 s / page | 0.8 s / page |
| Best for | Formatted text, formulas and charts | Lightweight OCR with boxes |
| Inputs | images | images |
| Languages | Chinese, English | English |
| Max pages per request | 100 | 500 |
| Tables | ✓ Yes | No |
| Handwriting | No | No |
| Math | ✓ Yes | No |
| Bounding boxes | ✓ Yes | ✓ Yes |
| Confidence | No | No |
| Forms | No | No |
| Multilingual | No | 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 GOT-OCR 2.0 if
- It outputs Markdown, LaTeX, TikZ, SMILES and kern, so formulas, tables, charts and sheet music come out as structured text.
- It supports region OCR guided by box coordinates or color, plus multi-page OCR with dynamic resolution.
- It targets English and Chinese documents, while Florence-2 gets Chinese output mostly wrong.
Choose Florence-2 if
- It is much smaller at 0.23B (base) or 0.77B (large), against 580M for GOT-OCR 2.0, and it is MIT licensed.
- OCR_WITH_REGION returns quadrilateral boxes per text line, which suits scene text and short labels.
- The same model also does captioning, detection and grounding, so one model can cover a pipeline that needs more than OCR.
Key differences
| GOT-OCR 2.0 | Florence-2 | |
|---|---|---|
| Licence | Code Apache-2.0; training data CC BY-NC 4.0 | MIT |
| Model size | 580M parameters | 0.23B (base), 0.77B (large) |
| Output formats | Plain text, Markdown, LaTeX, TikZ, SMILES, kern | Plain text and bounding boxes only; no Markdown, tables or LaTeX |
| Chinese text | Listed as a good fit (English and Chinese documents) | Output mostly wrong; vocabulary lacks CJK characters |
| Handwriting | Not published | A Hugging Face blog test found spelling errors and missing spaces on cursive handwriting |
| Hardware and setup | CUDA 11.8 and PyTorch 2.0.1 per the README; CPU users are pointed to a llama.cpp port | Self-hosted via Transformers; described as suited to lightweight edge OCR in English |
| Known failure | Loops characters midway through recognition (issue #226); outputs wrong text on blank images instead of an empty string (issue #235); format mode is much worse than plain mode (issue #317) | Spelling errors and missing spaces on French text; model card warns the continued-pretrained checkpoint used only 0.1B samples |
Sources
- GOT-OCR2.0 GitHub
- GOT-OCR 2.0 paper
- GitHub issue index
- Florence-2-large model card
- HF blog OCR test
- HF discussion Chinese very bad
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 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": "stepfun/got-ocr-2","fallbacks": ["microsoft/florence-2"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
GOT-OCR 2.0 alternativesFlorence-2 alternativesBest OCR models in 2026