OCRRouter

Donut vs Chunkr

Pick Chunkr if you need Markdown, HTML or JSON with bounding boxes from PDFs, slides, Word files and images, for example to chunk documents for RAG. Pick Donut if you have a narrow, fixed-layout document type such as receipts or forms and can fine-tune a model to output structured JSON. The main split is that Donut is a base model you must train, while Chunkr is a ready parsing service with a hosted API and a self-host option.

Donutnaver/donutChunkrchunkr/chunkr
TypeOpen sourceOCR API
Price1 credit / page14 credits / page
Per 1,000 pages$1.00$14.00
Typical speed1.0 s / page3.0 s / page
Best forReceipt and form fields after fine-tuningLayout-aware chunks for RAG
InputsimagesPDF, images, DOCX
LanguagesEnglish, Chinese, Japanese, KoreanMultilingual
Max pages per request1001,000
TablesNo✓ Yes
HandwritingNoNo
MathNo✓ Yes
Bounding boxesNo✓ Yes
ConfidenceNoNo
Forms✓ YesNo
MultilingualNo✓ 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 Donut if

  • Your documents are receipts, forms or another fixed layout, and you want fields extracted as JSON after fine-tuning.
  • You want an MIT-licensed model you can load through Hugging Face Transformers and use in closed products without a commercial license.
  • You need document classification, which Donut supports after fine-tuning.

Choose Chunkr if

  • You need full-document output as Markdown or HTML, which Donut is not meant for.
  • You want to start parsing PDFs, PPT, Word files and images without training anything.
  • You want a hosted API, or a Docker self-host with GPU, CPU-only and Mac ARM options.

Key differences

DonutChunkr
What it isOCR-free Swin encoder plus BART decoder model from NAVER Clova, published at ECCV 2022Document parsing service from Lumina AI with layout analysis, OCR and semantic chunking
Ready to useNo. The base checkpoint is pretrain-only and must be fine-tuned firstYes. Parses documents without training
LicenceMITAGPL-3.0 or commercial license; closed use needs the commercial license
How you run itSelf-host onlyCloud API or self-host via Docker
Output formatsJSON, plain textHTML, Markdown, JSON, bounding boxes
PriceFree, you pay for computeHosted credit-based plans, prices not verified; open-source version is free but you pay for compute and an LLM endpoint
Known limitsMisses tiny text in large images because of input resolution limits; Transformers v5 dropped image-to-text pipeline support, so use direct loading or v4Open-source version uses community models that are less accurate than the cloud's in-house models; no Excel support in the open-source version; self-host needs an external LLM endpoint

Sources

What each one is

Donut

OCR-free document understanding model. Fine-tuned for receipts, forms and document Q&A.

Donut 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": "naver/donut","fallbacks": ["chunkr/chunkr"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

Donut alternativesChunkr alternativesBest OCR models in 2026