Live from our benchmark engine

Model benchmarks

Standardized suites, temperature 0, blind judge. Read the methodology.

ModelSuiteQualityLatencyCost / runLast tested
GPT-5coding-v11003422 ms$0.001998Tue, 18 Aug 2026 10:59:12 GMT

GPT-5

quality · 2026-08-182026-08-18 · min 100 / max 100