What self-hosted RAG and LLM app platforms are available?

Examples include RAGFlow and Dify for AI applications, Qdrant and Weaviate for vector storage, and Langfuse for observability. They occupy different layers; this is not a verified combined stack. Check model endpoints and connector data flows.

Choose the layer you need

  • Separate application workflow and ingestion from vector storage, embeddings, inference, tracing and evaluation.
  • Test representative documents, retrieval quality, connector credentials, deletion and access control across the full data path.
  • Review model APIs, trace payloads, edition licensing and resource requirements independently for each selected component.

Avoid treating examples as a tested stack

The listings below illustrate several layers of an AI system. Their individual deployment paths do not certify interoperability or a common-provider deployment. Follow the integration documentation for the exact versions and measure your own ingestion, query and evaluation workloads.

Explore the reviewed listings

Illustrative candidates, shown alphabetically. Read each listing’s sources, edition limits and operating responsibilities; this is not a ranking or a feature-parity claim.

Sources and further guidance

Answer reviewed 2026-11-21. Check the exact edition, configuration and current vendor terms before deciding.

All 50 questions · Deployment guide · Build a shortlist