Marker vs Pixtral
Pick Marker when you need faithful PDF to Markdown conversion of books, papers with equations, or RAG-ready chunks, and you can live with its weight licence. Pick Pixtral only if you already run on the Mistral API and want to ask questions about charts and documents, since Mistral marks it deprecated and unmaintained. For new OCR work, Marker is the safer choice.
Markerdatalab/marker | Pixtralmistral/pixtral | |
|---|---|---|
| Type | OCR API | Vision model |
| Price | 14 credits / page | 2 credits / page |
| Per 1,000 pages | $14.00 | $2.00 |
| Typical speed | 2.5 s / page | 3.0 s / page |
| Best for | Books and research papers | Questions about document images |
| Inputs | PDF, images, DOCX | images |
| Languages | 90+ languages | Multilingual |
| Max pages per request | 1,000 | 500 |
| Tables | ✓ Yes | No |
| Handwriting | No | No |
| Math | ✓ Yes | No |
| Bounding boxes | ✓ Yes | No |
| Confidence | No | No |
| Forms | No | 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 Marker if
- It is built for PDF to Markdown and also outputs HTML, JSON block trees, LaTeX and RAG chunks.
- Datalab reports 83.7 on olmOCR-bench in vllm mode, and fast mode runs on CPU at 23.6 pages per second.
- It suits books and papers with equations, and you can run it on GPU, CPU or Apple Silicon.
Choose Pixtral if
- The open weights are Apache-2.0, with no revenue cap for commercial use.
- Mistral reports 90.7 ANLS on DocVQA, which fits question answering over documents and charts.
- The 128K token context allows many images per request, and the launch API price was $0.15 per million tokens.
Key differences
| Marker | Pixtral | |
|---|---|---|
| What it is | PDF to Markdown converter built on the Surya OCR models | General 12.4B vision language model (12B decoder plus 400M vision encoder) |
| Licence | Code Apache-2.0; weights under a modified AI Pubs Open Rail-M licence, free for research, personal use and startups under $5M | Apache-2.0 |
| Price | Free self-hosted within the weight licence; hosted API has about 5,000 free pages per month | Free to self-host; Mistral API launch price $0.15 per million input and output tokens |
| Output formats | Markdown, HTML, JSON, chunks, LaTeX, images | Plain text, Markdown, JSON |
| Benchmarks | olmOCR-bench 83.7 and OmniParseBench 85.9 in vllm mode, all Datalab's own numbers | DocVQA 90.7 ANLS, ChartQA 81.8, MMMU 52.5, per Mistral |
| Hardware | GPU, CPU or MPS; vllm mode needs an NVIDIA GPU; fast mode uses about 18 MB of ONNX weights on CPU | Not published |
| Known failure | CPU fast mode drops to 66.6 on olmOCR-bench | Reported accuracy drop on handwriting, small fonts and non-Latin scripts; a Q4 quantisation degrades small text OCR |
Sources
- Marker GitHub README
- Datalab pricing
- Mistral announcement
- Mistral Sept 2024 pricing
- Third-party deployment guide
What each one is
Marker
Hosted version of the open-source Marker and Surya models. Strong on books and papers.
Marker detailsPixtral
Mistral's open-weight vision models, Pixtral 12B and Pixtral Large. Mistral's API no longer serves them, so they run self-hosted.
Pixtral 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": ["mistral/pixtral"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Marker alternativesPixtral alternativesBest OCR models in 2026