The Postgres vector database that lets you store embeddings and run semantic search right next to your relational data.
Lantern is an open-source extension that turns Postgres into a fast vector database. Instead of running a separate vector store alongside your app, you keep embeddings, metadata, and application tables in one database — and query them with ordinary SQL. That dramatically simplifies the stack for any team already on Postgres building RAG, recommendations, or semantic search.
Who it's for: Engineering teams on Postgres who want vector search without standing up and maintaining a separate specialized database.
Adds fast approximate nearest-neighbor search to existing Postgres — no migration to a new system.
Combine keyword and semantic filters in a single query alongside your relational data.
MIT-licensed core you can run anywhere; managed option available for teams that want zero ops.
Built for real workloads with indexing tuned for billion-scale embedding sets.
If your app already lives in Postgres, Lantern is the lowest-friction way to add vector search. Skip it only if you need a purpose-built vector database with its own specialized feature set.