MinerU vs RolmOCR
Pick RolmOCR if you want an Apache-2.0 model that transcribes scanned or rotated pages to plain text on one GPU. Pick MinerU if you need Markdown, HTML, LaTeX or DOCX from mixed PDF and Office files, or a CPU-only start. MinerU scores 75.2 on olmOCR-Bench, but its VLM tiers drop merged table header rows and its licence terms need a check for commercial use.
MinerUopendatalab/mineru | RolmOCRreducto/rolmocr | |
|---|---|---|
| Type | Open source | Open source |
| Price | 2 credits / page | 2 credits / page |
| Per 1,000 pages | $2.00 | $2.00 |
| Typical speed | 2.0 s / page | 3.0 s / page |
| Best for | Academic PDFs | Fast open PDF to text |
| Inputs | PDF, images | PDF, images |
| Languages | 80+ languages | Not stated by the authors |
| Max pages per request | 500 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | ✓ Yes |
| Math | ✓ Yes | No |
| Bounding boxes | ✓ Yes | No |
| 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 MinerU if
- It accepts PDF, images, DOCX, PPTX, XLSX, EPUB, OFD and HTML in one tool, which suits mixed RAG corpora.
- It writes Markdown, HTML, LaTeX, DOCX, EPUB and JSON, so you get structure and not only text.
- Its basic ONNX tier runs on CPU with about 2 GB RAM, so you can start without a GPU.
Choose RolmOCR if
- Its Apache-2.0 weights can be used commercially without extra conditions.
- About 15% of its training data was rotated, so it is built for off-angle and skewed scans.
- It runs as a single model on one GPU, for example with vLLM, which keeps a self-hosted pipeline simple.
Key differences
| MinerU | RolmOCR | |
|---|---|---|
| What it is | Document parsing toolkit with CLI, API and WebUI, from CPU tier to a 1.2B VLM | Single OCR model fine-tuned from Qwen2.5-VL-7B, 8B parameters |
| Licence | MinerU Open Source License (Apache-2.0 based with extra conditions); VLM weights tagged AGPL-3.0 | Apache-2.0 |
| Output | Markdown, HTML, LaTeX, DOCX, EPUB, JSON | Plain text only; no layout bounding boxes |
| Hardware | Basic tier: CPU and 2 GB RAM. VLM tiers: 16 GB RAM and a GPU with 8 GB+ VRAM | Not published |
| Benchmark | 75.2 on olmOCR-Bench (MinerU 2.5.4); 45.9 mean on ParseBench | 0.663 overall on Needl.ai versus 0.770 for olmocr-2-7b |
| Known failure | Drops merged header rows in tables such as bank statements; repeated content can produce 300 KB figure blocks | Can hallucinate or drop content; omitted a table subtitle that olmOCR extracted |
| Price | Free, self-hosted (you pay for compute) | Free, self-hosted (you pay for compute) |
Sources
- MinerU GitHub README
- olmOCR-Bench table
- Issue #5554 table header rows
- Hugging Face model card
- Needl.ai benchmark
What each one is
MinerU
Open-source PDF to Markdown pipeline popular for academic and Chinese documents.
MinerU detailsRolmOCR
Reducto's open PDF-to-text model, built on Qwen2.5-VL-7B with the olmOCR recipe. Needs no PDF metadata and handles rotated pages.
RolmOCR 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": ["reducto/rolmocr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
MinerU alternativesRolmOCR alternativesBest OCR models in 2026