SQLite-vec

sqlite-vec is a SQLite extension providing vector search capabilities for embedded and local applications. Free and open-source (MIT). The extension adds virtual tables for vector storage and similarity search, enabling vector capabilities without separate database infrastructure. Supports float32, int8, and binary vector types with multiple distance functions (L2, cosine, inner product). The zero-dependency approach makes it ideal for desktop applications, mobile apps, edge deployments, and development environments. For text-to-SQL: local development and testing of RAG applications before deploying to production vector databases. Also suitable for applications requiring offline vector search capability. Limitations include SQLite’s single-writer constraint and lack of distributed operation. Comparison with Chroma: sqlite-vec is lower-level (extension vs full database) but integrates naturally with existing SQLite usage. Page should cover: installation and activation, virtual table creation, query patterns, supported vector types, integration with Python/JavaScript, comparison with Chroma embedded mode, and use cases for local/embedded vector search.