Acropolis

GPT-5 Codex

OpenAIGrade C. Evidence grade C of A to D: thin.ProprietaryReleased 23 Sept 2025
Compare with peers

Benchmark appearances

Best verified run per board, with confidence intervals and Pareto frontier status.

BenchmarkRankScoreCostStatusObserved
Terminal Bench

Epoch AI Benchmarking Hub · Codex CLI

#81 / 202
44.3%[39% to 49.6%]UnknownStandard15 Sept 2025
Terminal Bench

Epoch AI Benchmarking Hub · Codex CLI

#82 / 202
44.3%[39% to 49.6%]UnknownStandard4 Nov 2025
Terminal Bench

Epoch AI Benchmarking Hub · Terminus 2

#85 / 202
43.4%[37.8% to 49.1%]UnknownStandard15 Sept 2025
Terminal Bench

Epoch AI Benchmarking Hub · Terminus 2

#86 / 202
43.4%[37.7% to 49.1%]UnknownStandard31 Oct 2025
Terminal Bench

Epoch AI Benchmarking Hub · Mini-SWE-Agent

#102 / 202
41.3%[35.9% to 46.8%]UnknownStandard15 Sept 2025
Terminal Bench

Epoch AI Benchmarking Hub · Mini-SWE-Agent

#103 / 202
41.3%[35.8% to 46.8%]UnknownStandard3 Nov 2025