Back to the chart

HKUDS/RAG-Anything

All-in-one multimodal RAG framework built on LightRAG

Chart
Not on a chart right now
Stars
24.6k

Open on GitHubHomepage

What it is

RAG-Anything is a multimodal RAG framework that parses and queries documents containing text, images, tables, and mathematical equations in a single pipeline. Built on LightRAG, it handles document parsing, knowledge graph construction, and hybrid retrieval through one interface.

Who it is for

AI engineers and researchers working with mixed-content documents such as academic papers, technical documentation, and financial reports.

Why it is trending

It released v1.4.4 on October 28, 2026, and has accumulated 24,576 stars on GitHub for unifying multimodal document processing without separate specialized tools.

Main features

  • Universal document parsing for PDFs, Office files, and images
  • Dedicated analyzers for visual content, tables, and LaTeX equations
  • Multimodal knowledge graph construction with cross-modal relationships
  • Hybrid retrieval combining vector similarity search and graph traversal

Summarized on November 17, 2026 from the README and repository data. It can be wrong; check the source before relying on it. Report a wrong summary