bench

release

v2026.09

v2026.09 · suite v2026.09 · generated Sep 8, 2026

Releasesv2026.09v2026.09-smoke
Post on X

codex · gpt-6-astra · high leads at 77.8.

AgentOverallUX/UIFrontendBackendPlanning AuditBug FixTasteSignatureTokens inTokens outTurnsWall timeAPI-equiv. costScore / $Telemetry
01codex · gpt-6-astra · high77.8—55.6——100.0——5.5M111.7K767m 19s$65.501.2partial

Score vs. cost

0255075100$41.32$56.18$76.36$103.80scoreAPI-equivalent cost (log)codex · gpt-6-astra · high
codexPareto frontier