Announcing pg_vectorsearch: Scalable Open-Source Vector Search in PostgreSQL

pg_vectorsearch is a new open-source PostgreSQL extension in alpha that introduces the PRISM vector index, a partitioned and compressed alternative to pgvector’s graph-based HNSW index. PRISM clusters similar vectors, stores compact RaBitQ representations sequentially, and scans relevant partitions using SIMD-friendly access patterns, reducing memory requirements, index sizes, and build times while maintaining high recall. Reported benchmarks show that at 100 million vectors, PRISM achieved 6 ms median latency at 95% recall on a 128 GB machine, whereas HNSW required 512 GB and took far longer to build; at one million vectors, PRISM also substantially outperformed HNSW on low-memory machines. The extension is intended to let organizations retain vector search alongside relational application data in PostgreSQL, enabling SQL joins, transactional updates, metadata filtering, and hybrid keyword-vector search without operating a separate vector database. PRISM supports pgvector-compatible vector types and query operators, several distance functions, parallel index builds, and a single main query-tuning setting, though heavy update workloads currently require manual rebalancing and deleted space is not yet reused. The developers plan automatic rebalancing, filtered streaming search, production hardening, parallel queries, and other improvements before and after general availability.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.