dots.ocr vs RolmOCR
Pick dots.ocr when you need structure: layout boxes, tables as HTML, formulas as LaTeX, in one pass over multilingual PDFs. Pick RolmOCR when you only need plain text from scanned or rotated pages and want Apache-2.0 weights. The main split is output: dots.ocr returns layout and structure, RolmOCR returns plain text only and cannot give bounding boxes.
dots.ocrrednote/dots-ocr | RolmOCRreducto/rolmocr | |
|---|---|---|
| Type | Open source | Open source |
| Price | 2 credits / page | 2 credits / page |
| Per 1,000 pages | $2.00 | $2.00 |
| Typical speed | 1.5 s / page | 3.0 s / page |
| Best for | Multilingual document parsing | Fast open PDF to text |
| Inputs | PDF, images | PDF, images |
| Languages | 100+ languages | Not stated by the authors |
| Max pages per request | 500 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | ✓ Yes |
| Math | ✓ Yes | No |
| Bounding boxes | ✓ Yes | No |
| 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 dots.ocr if
- It returns JSON, Markdown, HTML, LaTeX and bounding boxes, so layout, tables and formulas come out of one model.
- At 3B parameters (1.7B LLM backbone) it is the smaller model, and the authors report 79.1 on olmOCR-bench for the original release and 83.9 for dots.mocr (v1.5).
- The dots.mocr-svg variant converts charts and diagrams to SVG, which RolmOCR does not offer.
Choose RolmOCR if
- Its Apache-2.0 licence is stated plainly, and the weights are usable commercially.
- About 15% of its training data was rotated, so it is built for off-angle and skewed scans.
- In the Needl.ai benchmark it logged 7 hallucination issues versus 11 for olmocr-2-7b, and it is a fit for plain text transcription on one GPU.
Key differences
| dots.ocr | RolmOCR | |
|---|---|---|
| Licence | MIT (the repository also includes a separate dots.ocr license agreement file) | Apache-2.0 |
| Model size | 3B (1.7B LLM backbone) | 8B (Qwen2.5-VL-7B base) |
| Output formats | JSON, Markdown, HTML, LaTeX, bounding boxes, SVG | Plain text only |
| Layout boxes | Returns layout boxes, categories and text | Cannot output bounding boxes |
| Benchmark | olmOCR-bench: 79.1 (dots.ocr), 83.9 (dots.mocr), reported by the authors | Needl.ai PDF extraction overall: 0.663 (independent), versus 0.770 for olmocr-2-7b |
| Known failure | Runs of ellipses or underscores can make output repeat endlessly; a table above 1000 grid cells raises an error and drops the page in Docling | Can hallucinate or drop content; one Reducto example omits a table subtitle that olmOCR extracted |
| Running it | Self-hosted via vLLM or Hugging Face Transformers; a single CUDA GPU with vLLM is the documented path; no VRAM figure stated | Self-hosted, for example with vLLM; free, you pay for compute; quantized versions have not been evaluated |
Sources
- dots.ocr GitHub README
- dots.ocr blog / limitations
- Docling issue #5146
- Hugging Face model card
- Needl.ai benchmark
What each one is
dots.ocr
Single vision model from rednote that does layout detection and content recognition together.
dots.ocr detailsRolmOCR
Reducto's open PDF-to-text model, built on Qwen2.5-VL-7B with the olmOCR recipe. Needs no PDF metadata and handles rotated pages.
RolmOCR 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": "rednote/dots-ocr","fallbacks": ["reducto/rolmocr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
dots.ocr alternativesRolmOCR alternativesBest OCR models in 2026