Marker vs Chandra
Pick Marker for self-hosted PDF to Markdown on modest hardware, including CPU or Apple Silicon, and for books and papers with equations or RAG chunking. Pick Chandra for multilingual scans and documents with tables, forms and math when you have an NVIDIA GPU or will use the hosted API. Chandra scores higher on olmOCR-bench (85.8 vs 83.7), but it needs a GPU and can hallucinate on hard pages.
Markerdatalab/marker | Chandradatalab/chandra | |
|---|---|---|
| Type | OCR API | Open source |
| Price | 14 credits / page | 2 credits / page |
| Per 1,000 pages | $14.00 | $2.00 |
| Typical speed | 2.5 s / page | 2.0 s / page |
| Best for | Books and research papers | Complex layouts and handwriting |
| Inputs | PDF, images, DOCX | PDF, images |
| Languages | 90+ languages | 40+ languages |
| Max pages per request | 1,000 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | ✓ Yes |
| Math | ✓ Yes | ✓ Yes |
| Bounding boxes | ✓ Yes | ✓ 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 Marker if
- It runs on GPU, CPU or Apple Silicon, and fast mode needs only about 18 MB of ONNX weights on CPU.
- Fast mode processes 23.6 pages per second on CPU, which suits bulk conversion without a GPU.
- It outputs Markdown, HTML, JSON block trees, RAG chunks, LaTeX and images, and its free threshold for startups is under $5M.
Choose Chandra if
- It scores 85.8 on olmOCR-bench, above Marker's 83.7 in vllm mode, and above dots.ocr 1.5 (83.9) and olmOCR 2 (82.4).
- It averages 77.8% across 43 languages, against 67.6% for Gemini 2.5 Flash, so it fits multilingual scans.
- It returns layout chunks with bounding boxes and labels per block, and the hosted API lists clear per-1,000-page prices of $1, $3 and $7.50.
Key differences
| Marker | Chandra | |
|---|---|---|
| Licence | Code Apache-2.0; weights under a modified AI Pubs Open Rail-M licence, free for research, personal use and startups under $5M | Code Apache-2.0; weights under modified OpenRAIL-M, free for research, personal use and companies under $2M funding or revenue |
| Hardware | GPU, CPU or MPS; vllm mode needs an NVIDIA GPU | Single NVIDIA GPU via vLLM; benchmarked on one H100 80GB, with presets for A100, L40S and RTX 4090/5090 |
| Hosted pricing | About 5,000 free pages per month; paid plans on Datalab's pricing page | $1 (fast), $3 (balanced) or $7.50 (accurate) per 1,000 pages; free tier of $20/month with a work email |
| Output formats | Markdown, HTML, JSON, chunks, LaTeX, images | Markdown, HTML, JSON, bounding boxes |
| olmOCR-bench | 83.7 in vllm mode; 66.6 in CPU fast mode (Datalab's own numbers) | 85.8 for Chandra 2 (authors' numbers) |
| Speed | 23.6 pages per second in CPU fast mode | About 1.44 pages per second on one H100 with vLLM; one reviewer measured 66 seconds per page on an RTX 4090 with the Hugging Face path |
| Known failure | Fast mode drops accuracy on olmOCR-bench to 66.6 | Token-loop hallucinations on 5-6 of about 60 Russian/Kazakh scans in GitHub issue #51; weak on Urdu (43.2%) and Sanskrit (51.1%); Old Scans category only 51.1 |
Sources
- Marker GitHub README
- Chandra GitHub README
- Datalab pricing
- GitHub issue #51 token loops
- Medium local test
What each one is
Marker
Hosted version of the open-source Marker and Surya models. Strong on books and papers.
Marker detailsChandra
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.
Use one, fall back to the other
{"model": "datalab/marker","fallbacks": ["datalab/chandra"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Marker alternativesChandra alternativesBest OCR models in 2026