GoKawiilParadeDB closes performance gap with PlanetScale's TIN search extension
ParadeDB published benchmark results showing it improved BM25-ranked text search throughput from 33 QPS to 81.9 QPS using the same StackExchange dataset and benchmarking harness PlanetScale used in its TIN launch post two weeks earlier. The company says the gains came from several optimization passes unrelated to document identifier handling, countering PlanetScale's claim that its speed advantage stemmed from using Postgres' internal ctid fields.
GoKawiil's interpretation of the reporting above, not reported fact.
The rapid turnaround suggests competitive pressure between Postgres search extensions is pushing faster iteration on performance, which could benefit developers choosing between these tools. ParadeDB's framing implies that architectural differences cited by PlanetScale may not be the primary driver of speed, a claim that invites further independent benchmarking to verify.
- ParadeDB roughly 2.5x'd its BM25 search throughput in two weeks, from 33 to 81.9 QPS
- The comparison used PlanetScale's own benchmarker tool and the same StackExchange dataset
- ParadeDB disputes PlanetScale's explanation that ctid-based document identification drives TIN's speed advantage
Source: paradedb.com, 2026-10-01
Published there as: “ParadeDB Search Performance Improvements”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.