Overview
pgvector lets you store and query vectors directly in PostgreSQL. It supports multiple distance metrics and indexing strategies, enabling hybrid queries with SQL, transactions, and filtering alongside vector similarity search.
Key features
- Native Postgres vector type and operators.
- HNSW and IVFFlat index support for approximate search.
- Broad client support across languages and easy deployment options.
Use cases
- RAG systems that benefit from SQL joins and strong consistency.
- Applications that require filtering and transactional guarantees with vector search.