Marker vs OpenOCR
Pick Marker if you want PDF to Markdown on your own machine, with HTML, JSON and RAG chunk output, a CPU mode, or a hosted API, and you are a research group, individual or startup under $5M. Pick OpenOCR if you need Apache-2.0 weights with no commercial limit and only handle Chinese and English documents. The main split is licence and language: Marker's weights restrict larger companies, while OpenOCR's models cover only two languages.
Markerdatalab/marker | OpenOCRfudan/openocr | |
|---|---|---|
| Type | OCR API | Open source |
| Price | 14 credits / page | 1 credit / page |
| Per 1,000 pages | $14.00 | $1.00 |
| Typical speed | 2.5 s / page | 0.5 s / page |
| Best for | Books and research papers | Lightweight OCR pipeline on CPU |
| Inputs | PDF, images, DOCX | PDF, images |
| Languages | 90+ languages | Chinese and English |
| Max pages per request | 1,000 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | No |
| Math | ✓ Yes | ✓ Yes |
| Bounding boxes | ✓ Yes | ✓ Yes |
| Confidence | No | No |
| Forms | No | No |
| Multilingual | ✓ Yes | No |
| 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
- You want a hosted option: Datalab sells an API with about 5,000 free pages per month, and you can also self-host.
- You need more output formats: Markdown, HTML, JSON block trees, chunks, LaTeX and images, including RAG chunking.
- You have no GPU: fast mode runs on CPU with about 18 MB of ONNX weights at 23.6 pages per second, though its olmOCR-bench score drops to 66.6.
Choose OpenOCR if
- You are a company above $5M and need free commercial use: code is Apache-2.0 and the pricing is free self-hosted.
- You parse Chinese and English academic PDFs with formulas: UniRec-0.1B recognizes plain text and single or multi-line formulas in one model.
- You want a small model: OpenDoc-0.1B has 0.1B parameters, reports 90.57 on OmniDocBench v1.5, and supports ONNX export.
Key differences
| Marker | OpenOCR | |
|---|---|---|
| Licence | Code Apache-2.0; weights under a modified AI Pubs Open Rail-M license (free for research, personal use and startups under $5M) | Apache-2.0 |
| How you run it | Self-host on GPU, CPU or Apple Silicon, or use Datalab's hosted API | Self-host as a Python library only; no hosted API |
| Price | Free self-hosted within the weight license; hosted API has about 5,000 free pages per month, paid plans on Datalab's pricing page | Free self-hosted; you pay for compute |
| Output formats | Markdown, HTML, JSON, chunks, LaTeX, images | Plain text, LaTeX, bounding boxes |
| Benchmark (self-reported) | 83.7 on olmOCR-bench and 85.9 on OmniParseBench in vllm mode; 66.6 on olmOCR-bench in CPU fast mode | 90.57 on OmniDocBench v1.5 (OpenDoc-0.1B) |
| Languages | Not published | Chinese and English only (OpenDoc and UniRec) |
| Known limitation | Weight license blocks commercial use for companies above $5M without a paid license | Document parsing depends on a separate layout model, PP-DocLayoutV2 from PaddlePaddle; no hardware requirements or throughput figures published |
Sources
What each one is
Marker
Hosted version of the open-source Marker and Surya models. Strong on books and papers.
Marker detailsOpenOCR
Fudan's open toolkit for text detection and recognition, formulas, tables and document parsing.
OpenOCR 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": ["fudan/openocr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Marker alternativesOpenOCR alternativesBest OCR models in 2026