MinerU

opendatalabReference management & readingListed

About this tool

An open-source document parser from OpenDataLab at Shanghai AI Laboratory. Convert paper PDFs, scans, Office files, EPUBs and web pages into Markdown, JSON and other formats while retaining equations, tables, images and reading order. Version 4.0 includes an official agent skill for reading local documents in page chunks with stable source locators. Documents are not uploaded to the cloud by default.

  • Feature 1: Convert equations to LaTeX, reconstruct tables and restore reading order.
  • Feature 2: Read PDFs, images, Office files, EPUB and HTML.
  • Feature 3: Export Markdown, JSON, HTML, LaTeX, DOCX and other formats.
  • Feature 4: Use the official agent skill for page-chunk reading and source locators.
  • Feature 5: Choose quality levels from flash to advanced; basic can run on a CPU.

Resource information

Maintainer
OpenDataLab, Shanghai AI Laboratory; official project.
Type
Skill
Category
Reference management & reading
Runs on
Agent Skills
Who it's for
Researchers and groups preparing large paper collections for agents, literature reviews or RAG, including equation-heavy papers and Chinese or English scans.
License at time of listing
MinerU Open Source License, based on Apache-2.0 with additional commercial licensing thresholds and attribution terms.
Source
GitHub