SQLite-Vector

A high-performance vector search extension for SQLite built on Faiss.

SQLite-Vector (sqlite-vss) brings production-grade similarity search to the local edge. It integrates the Faiss library directly into SQLite as a virtual table module, enabling BLOB-based vector storage and efficient k-nearest neighbor (k-NN) queries via the vss_search function. The extension supports IVFFlat indexing for sub-millisecond retrieval across datasets of 100,000+ embeddings. It is a lightweight, zero-config alternative to dedicated vector databases like Pinecone or Milvus, making it the ideal choice for local RAG applications and embedded AI tools.

https://github.com/asg017/sqlite-vss