MinerU vs docTR
Pick MinerU when you need whole documents turned into Markdown, HTML, LaTeX or DOCX, including Office files and academic papers. Pick docTR when you need word-level text with coordinates and plan to write your own extraction on top, for example for invoices and receipts. The main split is output: MinerU returns document structure, docTR returns words and lines only.
MinerUopendatalab/mineru | docTRmindee/doctr | |
|---|---|---|
| Type | Open source | Open source |
| Price | 2 credits / page | 1 credit / page |
| Per 1,000 pages | $2.00 | $1.00 |
| Typical speed | 2.0 s / page | 0.9 s / page |
| Best for | Academic PDFs | Word-level OCR with positions |
| Inputs | PDF, images | PDF, images |
| Languages | 80+ languages | Mainly Latin scripts |
| Max pages per request | 500 | 500 |
| Tables | ✓ Yes | No |
| Handwriting | No | No |
| Math | ✓ Yes | No |
| Bounding boxes | ✓ Yes | ✓ Yes |
| Confidence | No | ✓ Yes |
| 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 MinerU if
- It converts PDF, images, DOCX, PPTX, XLSX, EPUB, OFD and HTML to Markdown, HTML, LaTeX, DOCX, EPUB or JSON in one tool.
- The basic tier runs on CPU with about 2 GB RAM, and the VLM tier reaches 2.12 pages per second on one A100 with vLLM.
- It suits Chinese and English academic papers and mixed Office and PDF corpora for RAG.
Choose docTR if
- It gives a nested page, block, line and word structure with coordinates as JSON, plus bounding boxes.
- It is plain Apache-2.0, while MinerU uses a custom licence and its VLM weights are tagged AGPL-3.0.
- You can choose detection and recognition architectures and straight or rotated boxes, which suits Python teams training their own models.
Key differences
| MinerU | docTR | |
|---|---|---|
| What it is | Document parsing toolkit with CLI, API and WebUI | Two-stage detection and recognition OCR library in Python on PyTorch |
| Licence | MinerU Open Source License (Apache-2.0 based with extra conditions); MinerU2.5 VLM weights tagged AGPL-3.0 | Apache-2.0 |
| Output formats | Markdown, HTML, LaTeX, DOCX, EPUB, JSON | JSON, plain text, bounding boxes |
| Tables and formulas | Handled, but VLM tiers drop merged header rows such as bank statement account lines (issue #5554) | No table, formula or Markdown output; words and lines only |
| Hardware | Basic tier: CPU, 2 GB RAM. VLM tiers: 16 GB RAM and a GPU with 8 GB+ VRAM | CPU or NVIDIA GPU; GPU Docker images need CUDA 12.2 or newer on the host |
| Benchmark | olmOCR-Bench overall 75.2 (MinerU 2.5.4) | Not published |
| Known failure | VLM pipeline repeats content, producing 300 KB figure blocks (issue #5728) | Split-crop merging misjoins readings across repeated characters (issue #2229); cropping images gives poorer results (issue #2007) |
Sources
- MinerU GitHub README
- olmOCR-Bench table
- Issue #5554 table header rows
- docTR GitHub README
- Issue #2229
- Issue #2007
What each one is
MinerU
Open-source PDF to Markdown pipeline popular for academic and Chinese documents.
MinerU detailsdocTR
Open-source OCR library from Mindee with separate text detection and recognition models.
docTR 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": "opendatalab/mineru","fallbacks": ["mindee/doctr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
MinerU alternativesdocTR alternativesBest OCR models in 2026