Chandra vs Chunkr
Pick Chandra when you want a single OCR model for multilingual scans, tables, forms and math, with a published 85.8 on olmOCR-bench. Pick Chunkr when you need a parsing service for PDFs, slides and Word files with semantic chunking for RAG, and you can run your own LLM endpoint. The licences also split them: Chandra's weights are free for companies under $2M funding or revenue, while Chunkr's AGPL-3.0 needs a commercial license for closed products.
Chandradatalab/chandra | Chunkrchunkr/chunkr | |
|---|---|---|
| Type | Open source | OCR API |
| Price | 2 credits / page | 14 credits / page |
| Per 1,000 pages | $2.00 | $14.00 |
| Typical speed | 2.0 s / page | 3.0 s / page |
| Best for | Complex layouts and handwriting | Layout-aware chunks for RAG |
| Inputs | PDF, images | PDF, images, DOCX |
| Languages | 40+ languages | Multilingual |
| Max pages per request | 500 | 1,000 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | ✓ Yes | No |
| Math | ✓ Yes | ✓ Yes |
| Bounding boxes | ✓ Yes | ✓ Yes |
| Confidence | No | No |
| Forms | ✓ Yes | 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 Chandra if
- You work with scanned documents in many languages: Chandra 2 averages 77.8% across 43 languages.
- You want published numbers before you commit: 85.8 on olmOCR-bench, above dots.ocr 1.5 (83.9) and olmOCR 2 (82.4) in the authors' table.
- You want a hosted API with fixed per-page prices: $1, $3 or $7.50 per 1,000 pages, plus $20/month of free usage.
Choose Chunkr if
- You need to parse PDF, PPT, Word and image files in one service.
- You want semantic chunking and layout analysis for RAG over PDFs and slides.
- You want to self-host with Docker on GPU, CPU-only or Mac ARM, and point it at any OpenAI-compatible LLM endpoint.
Key differences
| Chandra | Chunkr | |
|---|---|---|
| Licence | Code Apache-2.0; weights modified OpenRAIL-M. Free for research, personal use and companies under $2M funding/revenue. Commercial license needed above that. | AGPL-3.0 or commercial license. Closed-source use needs the commercial license. |
| How you run it | Open weights on one NVIDIA GPU via vLLM or Hugging Face, or the hosted Datalab API. | Cloud API, or open-source self-host via Docker (GPU, CPU-only, Mac ARM). Self-host needs an external LLM endpoint. |
| Price | Self-hosted: free within license terms, you pay for compute. Hosted API: $1, $3 or $7.50 per 1,000 pages. | Hosted credit-based plans, price not published here. Open-source version is free, you pay for compute and an LLM endpoint. |
| Output formats | Markdown, HTML, JSON and bounding boxes, with layout labels per block. | HTML, Markdown, JSON and bounding boxes. |
| Input types | Not published | PDF, PPT, Word and images. Excel is not supported in the open-source version. |
| Benchmarks | olmOCR-bench overall 85.8 (Chandra 2). 77.8% average on 43 languages. | Not published |
| Known failure | Token-loop hallucinations reported on 5-6 of about 60 Russian/Kazakh scans. Weak on Urdu (43.2%) and Old Scans (51.1 on olmOCR-bench). | The open-source version uses community models, which are less accurate than the cloud's in-house models. |
Sources
What each one is
Chandra
Datalab's OCR model that keeps layout and handles forms, tables and handwriting.
Chandra detailsChunkr
Document parsing API that segments and chunks pages for RAG. Self-hostable version available.
Chunkr 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": "datalab/chandra","fallbacks": ["chunkr/chunkr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Chandra alternativesChunkr alternativesBest OCR models in 2026