MonkeyOCR vs GOT-OCR 2.0
Pick MonkeyOCR for English and Simplified Chinese PDFs such as academic papers with formulas and tables, but only for research or evaluation, because v1 weights need a written license for commercial use. Pick GOT-OCR 2.0 when you need a small 580M model that also reads sheet music, chemical structures and charts, or OCR on a chosen region of a page. GOT-OCR 2.0 publishes no benchmark scores and has open reports of looping output and wrong text on blank images.
MonkeyOCRmonkeyocr/monkeyocr | GOT-OCR 2.0stepfun/got-ocr-2 | |
|---|---|---|
| Type | Open source | Open source |
| Price | 2 credits / page | 1 credit / page |
| Per 1,000 pages | $2.00 | $1.00 |
| Typical speed | 1.5 s / page | 1.5 s / page |
| Best for | Chinese and English PDFs | Formatted text, formulas and charts |
| Inputs | PDF, images | images |
| Languages | Chinese, English | Chinese, English |
| Max pages per request | 500 | 100 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | No |
| Math | ✓ Yes | ✓ Yes |
| 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 MonkeyOCR if
- Its pro-3B model reports an OmniDocBench overall edit distance of 0.138 in English and 0.206 in Chinese, the best among the models the authors compared.
- It outputs Markdown, JSON, LaTeX, HTML and bounding boxes, plus a layout PDF with block coordinates.
- The pro-1.2B model is about 36% faster than pro-3B for about 1.6% lower accuracy, and quantized weights run on an 8GB RTX 4060.
Choose GOT-OCR 2.0 if
- It is a single 580M model that reads text, formulas, tables, charts, sheet music and geometry, and outputs formats such as TikZ, SMILES and kern.
- It supports region OCR guided by box coordinates or color, and multi-page OCR with dynamic resolution.
- It is merged into Hugging Face Transformers with batched inference, and the code is Apache-2.0, so setup is simpler than a custom LMDeploy or Docker stack.
Key differences
| MonkeyOCR | GOT-OCR 2.0 | |
|---|---|---|
| Licence | Code Apache-2.0; v1 weights research and non-commercial only, commercial use needs a written license | Code Apache-2.0; training data CC BY-NC 4.0 |
| Model size | 3B and 1.2B | 580M |
| How you run it | Self-hosted via LMDeploy or Docker; Docker on 20/30/40-series GPUs and V100 needs a patched image to avoid a shared-memory error | Self-hosted via Hugging Face Transformers; README lists CUDA 11.8 and PyTorch 2.0.1, with a llama.cpp port for CPU users |
| Output formats | Markdown, JSON, LaTeX, HTML, bounding boxes | Plain text, Markdown, LaTeX, TikZ, SMILES, kern |
| Benchmark | OmniDocBench pro-3B: edit distance 0.138 EN / 0.206 ZH, table TEDS 81.5 / 87.5 (reported by the authors) | Not published |
| Speed | Pro-3B about 0.49 pages/s on a 3090, under 1 page/s | Not published |
| Known failures | Not fully supported: photographed text, handwriting, Traditional Chinese and multilingual text | Reported issues: output loops characters midway, wrong text on images with no text, format mode much worse than plain OCR mode |
Sources
What each one is
MonkeyOCR
Open-source document parser that splits work into structure, recognition and relation steps.
MonkeyOCR detailsGOT-OCR 2.0
Unified OCR model that reads plain text, tables, formulas, sheet music and charts.
GOT-OCR 2.0 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": "monkeyocr/monkeyocr","fallbacks": ["stepfun/got-ocr-2"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
MonkeyOCR alternativesGOT-OCR 2.0 alternativesBest OCR models in 2026