PostgreSQL + pgvector

The industry standard for adding high performance vector similarity search to PostgreSQL.

Extend your existing PostgreSQL clusters (version 12+) with the pgvector extension to store and query high dimensional embeddings. It supports both HNSW and IVFFlat indexing strategies for sub-millisecond search latency. You can store up to 16,000 dimensions per vector while maintaining full ACID compliance and SQL flexibility. This setup eliminates the need for standalone vector stores: keep your relational tables and vector data in a single, battle tested environment.

https://github.com/pgvector/pgvector