Mathpix Convert vs Nanonets-OCR
Pick Mathpix Convert for math-heavy papers and textbooks, handwritten equations, and PDF to LaTeX or DOCX conversion through a hosted API. Pick Nanonets-OCR if you want to self-host a free model that turns printed business documents into Markdown for LLMs, with signatures, checkboxes and flowcharts tagged. The main split is STEM output you pay for per page versus a 3B open model you run yourself that was not trained on handwriting.
Mathpix Convertmathpix/convert | Nanonets-OCRnanonets/nanonets-ocr | |
|---|---|---|
| Type | OCR API | Open source |
| Price | 7 credits / page | 2 credits / page |
| Per 1,000 pages | $7.00 | $2.00 |
| Typical speed | 1.5 s / page | 1.8 s / page |
| Best for | Math and scientific papers | Markdown with semantic tags |
| Inputs | PDF, images | PDF, images |
| Languages | 30+ languages | Multilingual |
| Max pages per request | 500 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | ✓ Yes | No |
| Math | ✓ Yes | ✓ Yes |
| Bounding boxes | ✓ Yes | No |
| Confidence | ✓ Yes | No |
| Forms | No | ✓ Yes |
| Multilingual | No | ✓ 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 Mathpix Convert if
- You need LaTeX, Mathpix Markdown, DOCX or HTML output from whole PDFs with math, tables and chemistry.
- You have handwritten equations, which Nanonets-OCR was not trained on and can hallucinate on.
- You want a hosted API with no GPU to run, billed at $0.005 per PDF page or $0.002 per image.
Choose Nanonets-OCR if
- You want to self-host free weights from Hugging Face and pay only for compute.
- Your documents are business files with signatures, watermarks and checkboxes, which the model tags or converts to unicode symbols.
- You need flowcharts and org charts extracted as Mermaid code, or printed PDFs turned into Markdown for LLMs.
Key differences
| Mathpix Convert | Nanonets-OCR | |
|---|---|---|
| How you run it | Hosted API | Self-host the weights from Hugging Face |
| Licence | Proprietary | Not stated on the Nanonets-OCR2-3B model card (no licence tag); base model is Qwen2.5-VL-3B |
| Price | PDF $0.005 per page (0-1M), $0.0035 above 1M; images $0.002 each; one-time $19.99 API key activation fee | Free, you pay for compute |
| Output formats | LaTeX, Markdown, DOCX, HTML, JSON | Markdown, HTML, LaTeX, Mermaid |
| Handwriting | Handles handwritten equations | Not trained on handwriting; a handwriting test saw generation loops with CER up to 2100% and 69.1s per image on an L4 |
| Benchmarks | No published accuracy benchmark on general OCR leaderboards | olmOCR-bench overall 69.5, DocVQA 89.43, ChartQA 78.56 (OCR2-3B, reported by the authors) |
| Known failure | Images with more than 12 text rows may be billed at the PDF page rate | Trained on a fixed prompt; hallucination with other prompts is expected |
Sources
- Mathpix billing docs
- Mathpix OCR API
- Nanonets-OCR2-3B model card
- Nanonets-OCR-s announcement
- HF discussion fixed prompt
- Japanese handwriting OCR comparison
What each one is
Mathpix Convert
Specialist for math, science papers and handwritten formulas. Outputs LaTeX.
Mathpix Convert detailsNanonets-OCR
Open image-to-Markdown model that tags tables, equations, signatures, watermarks and checkboxes.
Nanonets-OCR 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": "mathpix/convert","fallbacks": ["nanonets/nanonets-ocr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Mathpix Convert alternativesNanonets-OCR alternativesBest OCR models in 2026