Memory & evals · Project

pgvector

pgvector adds vector storage and similarity search to PostgreSQL.

+ Follow

Builder facts

  • Database access. Version 0.8.7 supports PostgreSQL 13 or later. The extension is enabled per database and hosted PostgreSQL providers must make it available. Source ↗
  • License. The PostgreSQL License permits use and redistribution without a fee. Database hosting is separate. Source ↗
  • Index choice. Nearest-neighbor queries are exact by default. HNSW and IVFFlat indexes trade recall for speed. HNSW can be built before loading rows and takes more memory to build. Source ↗
  • Filtered retrieval. Approximate indexes apply filters after scanning, so selective filters can return fewer rows than requested. Iterative scans can search further up to a configured bound. Source ↗
  • Dimensions. The vector type stores up to 16,000 dimensions. HNSW and IVFFlat index up to 2,000 dimensions for vector or 4,000 for halfvec. Source ↗

Sources checked

Publisher update ·

pgvector 0.8.7 ↗

Fixed an IVFFlat index build buffer overflow and an error when the vector average aggregate had no matching rows.

Change timeline

0 stories

No verified releases in this timeline yet.