VectorChord

Variations

This product is available in the build below. Open it for the operating system, cloud and version detail, or read its deployment guide.

Overview

VectorChord adds the vchordrq vector index to PostgreSQL. It partitions embeddings into lists, compresses them with RaBitQ quantisation and reranks the closest candidates, so similarity search runs as an ordinary SQL query next to the rest of your data. It works on the pgvector vector type and distance operators, so existing pgvector tables and queries carry over, and pgvector's own HNSW and IVFFlat indexes remain available alongside it.

It suits teams building retrieval augmented generation, semantic search or recommendations who want their embeddings inside the PostgreSQL database they already run rather than in a separate vector service.

Common uses

• Retrieval augmented generation over embeddings stored in PostgreSQL
• Semantic and similarity search with SQL filters and joins
• Moving an existing pgvector workload to faster building indexes