MinerU

MinerU is an open-source document content extraction pipeline from OpenDataLab and the Shanghai AI Lab that converts PDFs and scanned documents into clean Markdown or JSON. It combines layout detection, OCR, formula recognition and table parsing in a single pipeline, and its approach was described in a September 2024 paper (arXiv 2409.18839). It has become a standard preprocessing step for RAG pipelines that need to ingest messy real-world documents.

One honest caveat on licensing: the project moved from AGPL-3.0 to a custom MinerU Open Source License that is based on Apache 2.0 but adds conditions, so check the current license text before commercial use. The team has since released MinerU2.5-Pro, a 1.2B vision-language variant that vendor-reported figures place at 95.69 on OmniDocBench (April 2026), pushing the line from pipeline tooling toward end-to-end document models.