Officially

From the publisher, with its history intact.

Databricks databricks.com

Lakebase Search: State-of-the-art full text and vector search for Postgres

Imported from official source

  • Lakebase Postgres now includes a built-in search engine (GA on AWS/Azure). Two extensions, lakebase_vector (ANN search) and lakebase_text (BM25), let you run semantic, keyword, and hybrid search directly in Postgres alongside operational data, eliminating the need for a separate search system + ETL pipeline.
  • It outperforms pgvector and dedicated search engines at scale. On a 100M-vector benchmark, Lakebase delivers 2x the throughput of the next-best system, 4x lower cost than cloud Postgres with pgvector, and 97% recall at 71ms P99 latency. It achieves this by decoupling storage from compute and using hierarchical IVF clustering + binary quantization (RaBitQ) so queries touch only the data they need.
  • The architecture is serverless and scales to zero. You pay for query usage, not data volume. Index builds are offloaded from the primary database, cold starts take ~1 second, and 100M vectors can be served on a single compute unit, making it purpose-built for the bursty retrieval patterns of AI agents.

    • Traditional OLTP systems weren't built for the search demands of AI agents. They require low-latency, high-accuracy retrieval across all your data and often execute massive parallel searches. Until now, solving this meant duct-taping a standalone search engine to your primary database with an ETL pipeline.

    But what if your OLTP database could just run the search workload efficiently?

    Today, we are bringing a fast and scalable search engine to Lakebase Postgres via two extensions: lakebase_vector (scalable approximate neighbor search) and lakebase_text (bm25 full-text search). Both extensions are generally available on AWS and Azure.

    This is an extract. The publication continues at the source.

    Read the original at the source: https://www.databricks.com/blog/lakebase-search-state-art-full-text-and-vector-search-postgres

    Officially imported this from Databricks’s own source and shows an extract. If you work there, claiming the profile and verifying the domain lets you choose to show the full text here.

    Provenance

    Organization
    Databricks — imported from official source
    Official source
    https://www.databricks.com/feed RSS
    Imported
    September 28, 2026 21:00
    Versions
    1 recorded
    Identity
    https://www.databricks.com/blog/lakebase-search-state-art-full-text-and-vector-search-p...

    Officially records where a publication came from, not whether it is true. Imported records are reproduced from an organization's own official source.