About Paradedb
ParadeDB is a search and analytics layer for Postgres. Its pg_search extension adds full text search, vector search, filters, facets and aggregates directly inside the database, so teams can avoid running a separate search system such as Elasticsearch and copying data into it.
The audience is developers and engineering teams who already store application data in Postgres and find queries becoming too slow. ParadeDB is shipped as an extension rather than a fork and works as a native index, created with a standard CREATE INDEX statement. It is built on Tantivy, a Rust search library inspired by Lucene, and offers tunable BM25 ranking, typo tolerance, highlighting, and phrase, fuzzy, proximity and regex queries. Because it stays inside Postgres, it keeps ACID transactions, joins, CTEs, row level security and the rest of standard SQL. The vendor publishes benchmark results against vanilla Postgres and Elasticsearch on its home page, and it describes the index as combining inverted, vector and columnar data structures.
The extension can be installed locally or on a self managed Postgres. A managed ParadeDB Cloud is available in private preview. The vendor does not publish prices on its site, and teams are directed to sign up or contact the company.