Post

Open Source pg_vectorsearch Brings More Scalable Vector Search to PostgreSQL

Tiger Data’s alpha pg_vectorsearch extension adds a partitioned, compressed PRISM vector index to PostgreSQL instead of relying on HNSW’s graph structure. It works with existing pgvector columns and requires PostgreSQL 18 or later.

The company reports substantially faster index construction on its 100-million-vector benchmark; treat those figures as vendor results. Incremental rebalancing is incomplete, and filtered vector searches are not yet supported. InfoQ reports that practitioners are asking for independent benchmark validation and discussing those limitations.