Pgvecto.Rs

Pgvecto.rs is a Postgres extension that adds vector similarity search to the database. It is written in Rust on top of the pgrx framework and introduces a vector type with a declared number of dimensions, so embeddings can be stored and queried alongside ordinary relational data. The documentation describes it as beta software. The documentation compares it with pgvector and lists several differences. It supports up to 65535 dimensions, where pgvector supports up to 2000. It adds binary vectors, 16 bit floating point and 8 bit integer data types. It uses a VBASE method so that filtered queries and joins return the full number of requested results. It dispatches SIMD instructions at runtime based on the machine's capabilities, and it manages index storage and memory separately from Postgres. Write ahead logging covers table data, while index support is still in progress. It suits developers who want vector search inside an existing Postgres database for retrieval, recommendations or semantic search. A Docker image is available for a quick start, after which the extension is enabled with a single SQL statement. The documentation carries a prominent warning recommending that users move to the vendor's newer VectorChord implementation, which it describes as offering better stability and performance, and it links a migration guide. No pricing is published for the extension.