MinerU 3: Document Extraction Demo

Open-source document extraction for PDF, DOCX, PPTX, XLSX, and images to Markdown and JSON.
If you found our project helpful, please give us a ⭐️ to support us!

Code Model Paper Homepage Download

Please select a file to upload (PDF, image, DOCX, PPTX, or XLSX)

Max convert pages

1 20

Backend

High-precision hybrid parsing, supports multiple languages.

Recognition Options:

If disabled, tables will be shown as images.

Enable table recognition

If disabled, inline formulas will not be detected or parsed.

Enable inline formula recognition

If disabled, image/chart blocks will keep layout positions but skip VLM image/chart analysis.

Enable image analysis

OCR Language

Select the OCR language for image-based PDFs and images.

Enable only if the result is extremely poor. Requires correct OCR language.

Force enable OCR

{}

Examples:

Conversion Status

Convert result