Infinity

Judgement

What it is

Building block

What it does

A database for LLM applications offering hybrid search across dense vectors, sparse vectors, multi-vector tensors and full text, from the makers of RAGFlow.

Adopt it when

You need hybrid retrieval with tensor reranking for RAG and want it in a single engine.

Watch out

Pre-1.0 with a small user base outside RAGFlow.

Ready for

Experimental

Runs as

Self-hosted server and library

Built with

C++ and Python

How established

Established: started 4 years ago, latest release v0.7.4.

What the license allows

You can build it into a product you sell, as long as you keep its copyright notice.

Judged on 2026-11-03 from the project's README and GitHub data. The license and age lines are worked out directly.

Built from

Follow it down: Infinity → LLVM project → runs on a microchip

How compilers were built, back to punched paper tape

Open-source alternatives to Infinity

1. 1
   Milvus milvus-io

   Milvus is a high-performance, cloud-native vector database built for scalable vector ANN search

   Building block Production-ready You need vector search over hundreds of millions of embeddings or more and have the people to run a distributed system.

   ★ 47k Apache-2.0 Go Active, pushed today
2. 2
   Qdrant qdrant

   Qdrant - High-performance, massive-scale Vector Database and Vector Search Engine for the next generation of AI. Also available in the cloud https://cloud.qdrant.io/

   Building block Production-ready You are building semantic search, RAG or recommendations and want a dedicated vector store that is simple to self-host.

   ★ 35k Apache-2.0 Rust Active, pushed today
3. 3
   Cognee topoteretes

   Cognee is the open-source AI memory platform for agents. Give your AI agents persistent long-term memory with small models for free

   Building block Usable You build agents that need long-term memory across sessions and want graph-based retrieval rather than plain vector RAG.

   ★ 33k Apache-2.0 Python Active, pushed today
4. 4
   Chroma chroma-core

   Search infrastructure for AI

   Building block Production-ready You are building retrieval for an LLM app and want something simple to start with locally.

   ★ 30k Apache-2.0 Rust Active, pushed 3 days ago
5. 5
   Zvec alibaba

   A lightweight, lightning-fast, in-process vector database

   Building block Usable You want SQLite-style embedded vector search inside an app or agent without a separate server.

   ★ 16k Apache-2.0 C++ Active, pushed today
6. 6
   LEANN StarTrail-org

   [MLsys2026 Best Paper]: https://arxiv.org/abs/2506.08276. RAG on Everything with LEANN. Enjoy 97% storage savings while running a fast, accurate, and 100% private RAG application on your personal device.

   Building block Experimental You want to search your own files, email or chat history with RAG on a laptop and storage is the constraint.

   ★ 14k MIT Python Active, pushed yesterday
7. 7
   LangChain4j langchain4j

   LangChain4j is an idiomatic, open-source Java library for building LLM-powered applications on the JVM. It offers a unified API over popular LLM providers and vector stores, and makes implementing tool calling (including MCP support), agents and RAG easy. It integrates seamlessly with enterprise Java frameworks like Quarkus and Spring Boot.

   Framework or platform Production-ready You are adding LLM features to a Java or Kotlin application and want to stay on the JVM.

   ★ 13k Apache-2.0 Java Active, pushed today
8. 8
   Txtai neuml

   💡 All-in-one AI framework for semantic search, LLM orchestration and language model workflows

   Framework or platform Usable You want one Python library to build semantic search and RAG locally with few external services.

   ★ 13k Apache-2.0 Python Active, pushed today

Commercial vector databases

1. 1
   Pinecone Pinecone Systems

   Pinecone is the trusted AI knowledge company. Its AI knowledge platform—Database, Nexus, and Marketplace—powers accurate, fast, cost-effective AI for 10,000+ customers and 1M developers worldwide.

   Building block Production-ready You need vector search at scale in production and do not want to run the infrastructure.

   Commercial