SmolDocling vs GOT-OCR 2.0
Pick SmolDocling only if you want to experiment with a tiny 256M document model on English pages, and note that its authors now point users to Granite-Docling. Pick GOT-OCR 2.0 if you need formulas, sheet music, chemical structures or region-specific OCR, and can accept English and Chinese only plus CC BY-NC training data. The main split is scope: SmolDocling is a small English page converter with structured tags, GOT-OCR 2.0 is a larger model that reads more content types.
SmolDoclinghuggingface/smoldocling | GOT-OCR 2.0stepfun/got-ocr-2 | |
|---|---|---|
| Type | Open source | Open source |
| Price | 1 credit / page | 1 credit / page |
| Per 1,000 pages | $1.00 | $1.00 |
| Typical speed | 1.0 s / page | 1.5 s / page |
| Best for | Very cheap page conversion | Formatted text, formulas and charts |
| Inputs | PDF, images | images |
| Languages | Mainly 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 SmolDocling if
- It has 256M parameters, so it runs on modest hardware, and it takes about 0.35 seconds per page on an A100 with vLLM.
- It outputs DocTags, Markdown, HTML, JSON, LaTeX and bounding boxes, which suits page conversion with location tags.
- It has published scores you can check: TEDS structure 0.82 on FinTabNet, full-page OCR F1 0.80 and OCRBench 338.
Choose GOT-OCR 2.0 if
- It reads formulas, sheet music, chemical structures and geometry, which SmolDocling does not list.
- It supports region OCR guided by box coordinates or color, plus multi-page OCR and dynamic resolution.
- It covers English and Chinese documents and is merged into Hugging Face Transformers with batched inference.
Key differences
| SmolDocling | GOT-OCR 2.0 | |
|---|---|---|
| Model size | 256M parameters | 580M parameters |
| Licence | Apache-2.0 per the model summary (metadata tag says CDLA-Permissive-2.0) | Code Apache-2.0; training data CC BY-NC 4.0 |
| Output formats | DocTags, Markdown, HTML, JSON, LaTeX, bounding boxes | Plain text, Markdown, LaTeX, TikZ, SMILES, kern |
| Languages | English only | English and Chinese |
| Published benchmarks | TEDS 0.82 (FinTabNet), page OCR F1 0.80, OCRBench 338 | Not published |
| Known failure | Endless token repetition and missing closing tags; output can loop on some images and become very slow | Output can loop characters midway; wrong text on blank images instead of an empty string; format mode is much worse than plain OCR mode |
| Status and hardware | Preview release, superseded by Granite-Docling; self-hosted, you pay for compute | CUDA 11.8 and PyTorch 2.0.1 per README, CPU users pointed to a llama.cpp port; self-hosted, you pay for compute |
Sources
- SmolDocling model card
- Docling issue #1515
- Granite-Docling comparison table
- GOT-OCR2.0 GitHub
- GOT-OCR 2.0 paper
What each one is
SmolDocling
Tiny 256M vision model from Hugging Face and IBM that outputs DocTags. Predecessor of Granite Docling.
SmolDocling 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": "huggingface/smoldocling","fallbacks": ["stepfun/got-ocr-2"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
SmolDocling alternativesGOT-OCR 2.0 alternativesBest OCR models in 2026