Claude vs MinerU
Pick Claude to read, summarize or question long reports and to extract fields to JSON with page-level citations. Pick MinerU to convert large PDF and Office corpora to Markdown, HTML, LaTeX or DOCX on your own hardware without paying per page. The main split is a hosted proprietary API that charges tokens per page versus a self-hosted toolkit that costs only compute.
Claudeanthropic/claude | MinerUopendatalab/mineru | |
|---|---|---|
| Type | Vision model | Open source |
| Price | 25 credits / page | 2 credits / page |
| Per 1,000 pages | $25.00 | $2.00 |
| Typical speed | 2.6 s / page | 2.0 s / page |
| Best for | Long dense documents | Academic PDFs |
| Inputs | PDF, images | PDF, images |
| Languages | 50+ languages | 80+ languages |
| Max pages per request | 100 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | ✓ Yes | No |
| Math | ✓ Yes | ✓ Yes |
| Bounding boxes | No | ✓ Yes |
| 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 Claude if
- Native PDF input returns page-level citations, which suits document question answering.
- It takes up to 600 pages per request on 1M-context models (100 pages otherwise) with a 32 MB payload, so long reports go in whole.
- Each page is read as text and as a rendered image, so charts and scans are visible to the model, and it can return JSON with reasoning.
Choose MinerU if
- It is free to self-host, and the basic ONNX tier runs on CPU with 2 GB RAM, so you can batch-convert with no GPU and no per-page fee.
- It accepts PDF, images, DOCX, PPTX, XLSX, EPUB, OFD and HTML and renders to Markdown, HTML, LaTeX, DOCX, EPUB or JSON.
- It has published independent scores (olmOCR-Bench overall 75.2 for MinerU 2.5.4), and the VLM tier reaches 2.12 pages per second on one A100 with vLLM.
Key differences
| Claude | MinerU | |
|---|---|---|
| How you run it | Hosted API | Self-host as CLI, API or WebUI |
| Licence | Proprietary | MinerU Open Source License (Apache-2.0 based with extra conditions); MinerU2.5 VLM weights tagged AGPL-3.0 |
| Price | Tokens per model, no extra PDF fee; a PDF page typically uses 1,500 to 3,000 tokens plus image tokens | Free; you pay for compute |
| Input and output | PDFs and images in; Markdown, JSON or plain text out | PDF, images, Office files, EPUB, OFD, HTML in; Markdown, HTML, LaTeX, DOCX, EPUB, JSON out |
| Hardware and limits | Up to 600 pages per request on 1M-context models (100 otherwise), 32 MB payload; encrypted PDFs not supported | Basic tier: CPU, 2 GB RAM. VLM tiers: 16 GB RAM and a GPU with 8 GB+ VRAM |
| Benchmarks | No published OmniDocBench score for any Claude model in the leaderboards checked | olmOCR-Bench overall 75.2 (MinerU 2.5.4); ParseBench mean 45.9 (MinerU2.5) |
| Known failure | Bulk OCR is costly per page; no bounding boxes | VLM tiers drop merged header rows in tables such as bank statement account lines; VLM pipeline can generate repeated content, including 300 KB figure blocks; full-width Chinese punctuation is converted to half-width with no switch |
Sources
- Claude PDF support docs
- MinerU GitHub README
- olmOCR-Bench table
- Issue #5554 table header rows
- Issue #5728 repetition
What each one is
MinerU
Open-source PDF to Markdown pipeline popular for academic and Chinese documents.
MinerU detailsTry both on your own file
Run the file once with each model and compare the output.
PDF, PNG, JPG, WebP, HEIC. Up to 20 MB each, 10 files.No file?
Use one, fall back to the other
{"model": "anthropic/claude","fallbacks": ["opendatalab/mineru"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Claude vs DoclingClaude vs Unlimited OCRLlamaParse vs MinerUGPT vs MinerUMinerU vs DoclingMinerU vs InternVL3.5Claude vs Reducto ParseClaude vs docTR
Claude alternativesMinerU alternativesBest OCR models in 2026