HKUDS/RAG-Anything

Health v2 · maintenance only

Adoption not part of the score

23584 stars · 2748 forks observed · 2026-10-16

What it is AI-extracted, prompt v1, taxonomy v1, 2026-10-12, confidence 0.97

RAG-Anything is a Python-based, all‑in‑one Retrieval‑Augmented Generation framework that supports text, images, tables, and math. It provides unified multimodal retrieval, hybrid search, and integration with large language models.

Use cases

• retrieve relevant information from multimodal documents such as PDFs with images and tables
• augment LLM responses with visual content like charts, diagrams, or equations
• build a RAG pipeline that can search across text, images, and structured data in a knowledge base
• enable cross‑modal question answering where evidence spans multiple modalities
• create chat assistants that cite both textual and visual sources

When to choose

• You need a single framework that handles text, images, tables, and mathematical expressions in RAG pipelines.
• Your project is Python‑centric and you want easy integration with popular LLMs.
• You require multimodal retrieval and hybrid search (semantic + structural) out of the box.
• You prefer an actively maintained open‑source solution with a large community.

When to avoid

• Your use case is limited to pure text retrieval and you want a lightweight, text‑only library.
• You need a high‑performance C/C++ implementation for extreme low‑latency environments.
• You are constrained to non‑Python runtimes or platforms without Python support.
• You require a fully managed SaaS service rather than self‑hosted code.

Facets

framework · maturity active

search-engine vector-database llm-inference artificial-intelligence large-language-models python windows cross-platform multimodal all-in-one open-source mit-license rag knowledge-graph retrieval-augmented-generation linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="HKUDS/RAG-Anything")