RAGFlow
InfiniFlowOpen-source RAG engine built around deep document understanding - template-based chunking and layout-aware parsing of complex documents (PDFs, tables, figures, scans) to produce grounded, citation-backed answers. Unlike general orchestration frameworks, it is a batteries-included RAG application/server with strong document parsing and visible source citations, recently extended with agent capabilities.
Apache-2.0. ~83k stars. Deployed via Docker Compose (no first-party PyPI/npm). Commercial managed cloud / enterprise edition exists; the self-hosted engine is fully open.
Openness
5 high confidence- license
- Apache-2.0(OSI)
- source
- public
- self-host
- primary(Docker)
- commercial
- cloud/enterprise(separate)
- core-gated
- ungated
Apache-2.0 self-hostable engine; cloud is the commercial tier.
- https://github.com/infiniflow/ragflow recorded 2026-06-18
Apache-2.0, ~83k stars, RAG engine
Adoption
3 medium confidence~83k stars; Docker-distributed with no download metric retrieved - stars_fallback (capped at 3).
- https://ragflow.io recorded 2026-06-18
official site, InfiniFlow, self-host + cloud
Capability
4 medium confidenceOpinionated, document-centric RAG application; deeper parsing than general frameworks, narrower scope.
- https://github.com/infiniflow/ragflow recorded 2026-06-18
feature set: deep doc understanding, citations, hybrid retrieval
Unchanged since 2026-07-30 (last edited, not re-checked)