PlanetScale launches TIN: fast full-text search for Postgres
Original: Tin: full-text search for Postgres
Why This Matters
Full-text search parity with dedicated search engines is a major gap Postgres users work around daily.
PlanetScale released TIN (Text INdex), a full-text search extension for Postgres, on September 16, 2026. Available immediately as a GA release for all Postgres and Neki databases, TIN supports Boolean expressions, phrase queries, fuzzy/wildcard/regex matching, BM25 scoring, and continuous updates with correct MVCC transaction visibility.
PlanetScale has shipped TIN, a new full-text search index extension for Postgres, available now in general availability for both its standard Postgres and sharded Neki databases. The syntax is minimal: one CREATE INDEX call using the 'tin' access method, and queries use the ==> operator.
The team says they evaluated at least three existing Postgres text-search indexes and found none satisfied all their requirements simultaneously—specifically, correct transaction visibility under MVCC, support for COUNT(*) and BM25-ranked top-k queries, phrase and span queries, fuzzy/wildcard/regex term matching, case and accent folding, and live updates during ongoing query workloads.
Benchmarks were run against an 85 GB Stack Exchange export with 150 million documents, with additional testing on Wikipedia, 2.3 TB of Reddit comments, and a mixed 797 GB corpus of research papers, legal documents, books, and Enron emails. The team tested conjunction, disjunction, and phrase query patterns, both with and without concurrent writes. Performance claims are described as 'mind-blowingly fast,' though exact QPS figures from the article content provided are pending the full benchmark tables. Key internal engineering decisions include 48-bit document identifiers, work elision, vectorization, and a segment-based merge architecture to handle MVCC correctly at scale.