Claude vs Chandra
Pick Claude when you need to read, question or extract structured data from long PDFs, with page-level citations. Pick Chandra when you convert many scanned pages to Markdown, HTML or JSON with bounding boxes, or when you want to self-host. The main split is that Claude bills by tokens at 1,500 to 3,000 per page and gives no bounding boxes, while Chandra is a dedicated OCR model with per-page pricing and layout output.
Claudeanthropic/claude | Chandradatalab/chandra | |
|---|---|---|
| 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 | 2.0 s / page |
| Best for | Long dense documents | Complex layouts and handwriting |
| Inputs | PDF, images | PDF, images |
| Languages | 50+ languages | 40+ languages |
| Max pages per request | 100 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | ✓ Yes | ✓ Yes |
| Math | ✓ Yes | ✓ Yes |
| Bounding boxes | No | ✓ Yes |
| Confidence | No | No |
| Forms | No | ✓ Yes |
| 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
- You need document question answering or summaries with page-level citations from native PDF input.
- You want extraction to JSON with reasoning, in one request of up to 600 pages on 1M-context models (100 pages otherwise, 32 MB payload).
- You need the model to see charts and scans, since each page is read as both extracted text and a rendered image.
Choose Chandra if
- You run high-volume OCR and want fixed per-page pricing: $1, $3 or $7.50 per 1,000 pages on the hosted API, or free self-hosting within the licence terms.
- You need layout chunks with bounding boxes and labels per block, plus Markdown, HTML or JSON output.
- You process multilingual scans: Chandra 2 averages 77.8% across 43 languages in the authors' benchmark, and it targets tables, forms and math.
Key differences
| Claude | Chandra | |
|---|---|---|
| How you run it | Hosted API only | Self-host the open weights (vLLM or Hugging Face) or use the hosted Datalab API |
| Licence | Proprietary | Code Apache-2.0; weights modified OpenRAIL-M, free for research, personal use and companies under $2M funding/revenue |
| Price | Token-based, no extra PDF fee; a PDF page uses 1,500 to 3,000 tokens plus image tokens | Hosted: $1, $3 or $7.50 per 1,000 pages, $20/month free tier; self-hosted: free within licence terms, you pay for compute |
| Output formats | Markdown, JSON, plain text | Markdown, HTML, JSON, bounding boxes |
| Benchmark | No published OmniDocBench score | Chandra 2: 85.8 on olmOCR-bench (authors' figure) |
| Input limits | Up to 600 pages per request on 1M-context models, 100 otherwise; 32 MB; no password-protected PDFs | Not published |
| Known failure | Costly for bulk OCR; encrypted PDFs not supported | Token-loop hallucinations on some Russian/Kazakh scans (GitHub issue #51); weak on Urdu (43.2%), Sanskrit (51.1%) and old scans |
Sources
- Claude PDF support docs
- Claude PDF page limits
- Chandra GitHub README
- Datalab pricing
- GitHub issue #51 token loops
What each one is
Chandra
Datalab's OCR model that keeps layout and handles forms, tables and handwriting.
Chandra 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": ["datalab/chandra"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Claude vs SuryaClaude vs olmOCRLlamaParse vs ChandraGPT vs ChandraClaude vs DonutClaude vs Nanonets-OCRClaude vs Qianfan OCRClaude vs Phi-4-multimodal
Claude alternativesChandra alternativesBest OCR models in 2026