PlanetScale's TIN beats Postgres GIN full-text search: 212ms vs 288s at p99

DanielLockyer · x · 2026-09-18

PlanetScale has introduced TIN, a full-text search extension for Postgres, and a developer benchmark is turning heads: the same query clocks in at a p99 of 212 milliseconds on TIN versus a staggering 288 seconds with Postgres's built-in GIN index — a three-orders-of-magnitude gap.

TIN works with complicated WHERE clauses, replication, and backups, while maintaining correct transaction visibility. Given how widely GIN has been used for Postgres full-text search, the result suggests many production setups may be worth revisiting.

Original post →

More from Infra

Infra channel →