Claude vs dots.ocr
Pick Claude for reading long reports, extracting to JSON with reasoning, and answering questions with page citations. Pick dots.ocr for high-volume parsing of multilingual PDFs into Markdown with layout boxes, since it is MIT-licensed code you run on your own GPU and you pay only for compute. The main split is a hosted per-token API versus a self-hosted 3B parser that returns bounding boxes.
Claudeanthropic/claude | dots.ocrrednote/dots-ocr | |
|---|---|---|
| Type | Vision model | Open source |
| Price | 25 credits / page | 2 credits / page |
| Per 1,000 pages | $25.00 | $2.00 |
| Typical speed | 2.6 s / page | 1.5 s / page |
| Best for | Long dense documents | Multilingual document parsing |
| Inputs | PDF, images | PDF, images |
| Languages | 50+ languages | 100+ languages |
| Max pages per request | 100 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | ✓ Yes | No |
| Math | ✓ Yes | ✓ Yes |
| Bounding boxes | No | ✓ Yes |
| Confidence | No | No |
| Forms | No | No |
| Multilingual | ✓ Yes | ✓ Yes |
| 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 Claude if
- It takes PDFs natively and gives page-level citations in the API, which suits document question answering.
- It accepts up to 600 pages per request on 1M-context models (100 pages otherwise) with a 32 MB payload.
- It reads each page as both extracted text and a rendered image, so charts and scans are visible to the model.
Choose dots.ocr if
- It returns layout boxes, categories and text in one pass, with tables as HTML and formulas as LaTeX, which Claude does not offer for pipelines that need bounding boxes.
- It is free to run on your own hardware under MIT-licensed code, so bulk jobs cost compute rather than 1,500 to 3,000 tokens per page.
- It reports 83.9 on olmOCR-bench for dots.mocr (v1.5), up from 79.1 for the original, and the dots.mocr-svg variant converts charts and diagrams to SVG.
Key differences
| Claude | dots.ocr | |
|---|---|---|
| How you run it | Hosted API | Self-hosted via vLLM or Hugging Face Transformers; a single CUDA GPU with vLLM is the documented path |
| Licence | Proprietary | MIT code, plus a separate dots.ocr license agreement file in the repository |
| Price | Token-based per model, no extra PDF fee; a PDF page typically uses 1,500 to 3,000 tokens plus image tokens | Free, you pay for compute |
| Output formats | Markdown, JSON, plain text | JSON, Markdown, HTML, LaTeX, bounding boxes, SVG |
| Benchmarks | No published OmniDocBench score for any Claude model in the leaderboards checked | olmOCR-bench 79.1 (dots.ocr) and 83.9 (dots.mocr); OmniDocBench v1.5 text edit distance 0.031 for dots.mocr (lower is better), all reported by the authors |
| Known failure | Password-protected or encrypted PDFs are not supported | Runs of ellipses or underscores can make output repeat endlessly; complex tables and formulas remain difficult; in Docling, a table above 1000 grid cells raises an error and drops the whole page |
| Page limits | Up to 600 pages per request on 1M-context models, 100 otherwise; 32 MB payload | Not published |
Sources
- Claude PDF support docs
- Claude PDF page limits
- dots.ocr GitHub README
- dots.ocr blog / limitations
- Docling issue #5146
What each one is
dots.ocr
Single vision model from rednote that does layout detection and content recognition together.
dots.ocr detailsTry both on your own file
Run the file once with each model and compare the output.
PDF, PNG, JPG, WebP, HEIC. Up to 20 MB each, 10 files.No file?
Use one, fall back to the other
{"model": "anthropic/claude","fallbacks": ["rednote/dots-ocr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Claude vs Nanonets-OCRClaude vs SuryaLlamaParse vs dots.ocrGPT vs dots.ocrJina OCR vs dots.ocrClaude vs Qianfan OCRdots.ocr vs SmolDoclingPaddleOCR-VL vs dots.ocr
Claude alternativesdots.ocr alternativesBest OCR models in 2026