MinerU
by OpenDataLab
Open-source PDF extraction toolkit with high-quality conversion to Markdown and structured formats
Overview
MinerU is a comprehensive PDF extraction toolkit from OpenDataLab designed for converting documents into structured, machine-readable formats. It combines multiple models for OCR, layout analysis, and content understanding into a unified pipeline.
The 2.5 release (1.2B parameters) significantly improved table extraction, formula recognition, and multi-column layout handling. MinerU outputs clean Markdown with preserved document structure, making it ideal for RAG pipelines and document processing workflows.
It's positioned as an open-source alternative to commercial document AI services, with particular strength in academic and technical documents.
Strengths
- High-quality PDF to Markdown conversion
- Strong table and formula extraction
- Multi-column layout support
- Active development with frequent updates
- Good for RAG preprocessing
Limitations
- Requires GPU for optimal performance
- Processing speed varies with document complexity
- May struggle with heavily stylized documents
Best Use Cases
- PDF to Markdown conversion
- RAG pipeline preprocessing
- Academic paper processing
- Technical documentation extraction