OCRRouter

dots.ocr vs Chunkr

Pick dots.ocr if you want a 3B model on your own GPU that turns multilingual PDFs into Markdown, HTML tables and LaTeX formulas under an MIT licence. Pick Chunkr if you need a ready parsing service, with a hosted API or a self-host option, that accepts PDF, PPT, Word and images and does semantic chunking for RAG. The main split is a single open model you run yourself versus a parsing pipeline that needs an LLM endpoint and, for closed-source use, a commercial licence.

dots.ocrrednote/dots-ocrChunkrchunkr/chunkr
TypeOpen sourceOCR API
Price2 credits / page14 credits / page
Per 1,000 pages$2.00$14.00
Typical speed1.5 s / page3.0 s / page
Best forMultilingual document parsingLayout-aware chunks for RAG
InputsPDF, imagesPDF, images, DOCX
Languages100+ languagesMultilingual
Max pages per request5001,000
Tables✓ Yes✓ Yes
HandwritingNoNo
Math✓ Yes✓ Yes
Bounding boxes✓ Yes✓ Yes
ConfidenceNoNo
FormsNoNo
Multilingual✓ Yes✓ Yes
On OCRRouterAvailableAvailable

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

  • The code is MIT-licensed, so you can use it in a closed product without buying a commercial licence.
  • It has published scores: 79.1 on olmOCR-bench for the original and 83.9 for dots.mocr (1.5), as reported by the authors.
  • One model returns layout boxes, categories and text, with tables as HTML and formulas as LaTeX, and dots.mocr-svg also converts charts and diagrams to SVG.

Choose Chunkr if

  • It accepts PDF, PPT, Word and images, and it does semantic chunking, which suits RAG over PDFs and slides.
  • You can use the hosted cloud API or self-host the open-source version with Docker on GPU, CPU-only or Mac ARM.
  • The self-host version works with any OpenAI-compatible LLM endpoint, so you can choose the model behind it.

Key differences

dots.ocrChunkr
LicenceMIT (the repository also includes a separate dots.ocr licence agreement file)AGPL-3.0 or commercial licence; closed use needs the commercial one
How you run itSelf-host only, with vLLM or Hugging Face Transformers; a single CUDA GPU with vLLM is the documented pathCloud API or self-host via Docker (GPU, CPU-only, Mac ARM); self-host still needs an external LLM endpoint
PriceFree, you pay for computeHosted credit-based plans, prices not published; open-source version is free, you pay for compute and an LLM endpoint
Output formatsJSON, Markdown, HTML, LaTeX, bounding boxes, SVGHTML, Markdown, JSON, bounding boxes
Input typesNot publishedPDF, PPT, Word and images; no Excel in the open-source version
BenchmarksolmOCR-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 from the authorsNot published
Known failureRuns of ellipses or underscores can make output repeat endlessly; complex tables and formulas stay difficult; in Docling, a table above 1000 grid cells raises an error and drops the whole pageThe open-source version uses community models that are less accurate than the cloud's in-house models

Sources

What each one is

dots.ocr

Single vision model from rednote that does layout detection and content recognition together.

dots.ocr details

Chunkr

Document parsing API that segments and chunks pages for RAG. Self-hostable version available.

Chunkr details

Try 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": "rednote/dots-ocr","fallbacks": ["chunkr/chunkr"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

dots.ocr alternativesChunkr alternativesBest OCR models in 2026