# Lakebase Search: vector and BM25 on Neon

DevFeed: [Lakebase Search: vector and BM25 on Neon](<https://devfeed.tech/articles/lakebase-search-vector-and-bm25-on-neon-5484.md>)

Original publisher: [Read original article](<https://neon.com/blog/lakebase-search-on-neon>)

Author: Savannah Longoria

Published: 2026-07-02T08:49:00Z

Content type: release

Language: en

Sources: [Blog -- Neon Docs](<https://devfeed.tech/sources/blog-neon-docs.md>)

Topics: [Databases](<https://devfeed.tech/topics/databases.md>), [benchmarking](<https://devfeed.tech/topics/benchmarking.md>), [SQL](<https://devfeed.tech/topics/sql.md>)

Tags: [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [company](<https://devfeed.tech/tags/company.md>), [database](<https://devfeed.tech/tags/database.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [search](<https://devfeed.tech/tags/search.md>), [sql](<https://devfeed.tech/tags/sql.md>), [vector](<https://devfeed.tech/tags/vector.md>)

## AI overview

Lakebase Search on Neon provides hybrid vector and full-text retrieval through two Postgres extensions. The article explains its lakebase storage-compute architecture, contrasts IVF and RaBitQ with HNSW, and describes BM25 with top-K pushdown as an alternative to GIN-based keyword search.

## Source excerpt

Today we're shipping the beta of Lakebase Search on Neon: hybrid vector + full-text retrieval via two Postgres extensions, lakebase_vector and lakebase_text. This post is the story of why and how we built it: IVF + RaBitQ instead of HNSW, BM25 with top-K pushdown instead of GIN on tsvector, and indexes that live on object storage so they survive scale-to-zero and branch instantly.