RAGFlow

lastmod 2026-08-13created 2026-08-131 edits

Open-source RAG engine based on OCR and document parsing

An open-source RAG (retrieval-augmented generation) engine based on deep document understanding

Adds a streamlined RAG workflow to businesses of any size, and provides LLMs with factual question-answering capabilities backed by well-grounded citations from data in various complex formats

Key features

• "Quality in, quality out" : Knowledge extraction from unstructured data with complex formats, based on deep document understanding. Enables "finding a needle in a data haystack" across unlimited tokens
• Template-based chunking : Intelligent and easy to explain. Choose from a variety of template options
• Reduced hallucinations through grounded citations : Visualizes text chunking to allow human intervention. Supports well-grounded answers by quickly checking key references and traceable citations
• Compatibility with heterogeneous data sources : Supports Word, slides, Excel, txt, images, scanned copies, structured data, web pages, and more
• Automated and easy RAG workflow
  • Streamlined RAG orchestration for both individuals and large enterprises
  • Configurable LLMs, including embedding models
  • Multiple recall combined with fused reranking
  • Intuitive APIs for seamless integration with businesses

See also

• OCR
• OpenDocuments – An open-source RAG platform for searching scattered organizational documents in natural language

Favorite site

• Github - infiniflow/ragflow
• RAGFlow - Open-source RAG engine based on OCR and document parsing | GeekNews