Skip to content

OpenAI

GPT-4 Turbo (Apr 2024)

Announced
Access
Public

Benchmark scores

TestScoreBest on the trackerHow it was measured
FrontierMath T1–3hard maths problems0.7%Claude Fable 5.2 95%Independentdefault settings, turbo endpoint (gpt-4-turbo-2024-04-09)Run 27 Aug 2026
GPQA DiamondPhD-level science questionsPhD experts: 70%Saturated47%Gemini 4 Argon 96%Independentdefault settings, turbo endpoint (gpt-4-turbo-2024-04-09)Run 27 Feb 2025
OTIS Mock AIMEcompetition mathsSaturated6.7%GPT-5.5 100%Independentdefault settings, turbo endpoint (gpt-4-turbo-2024-04-09)Run 27 Feb 2025
METR time horizontask length an AI agent finishes half the timeNo longer updated3.7 minClaude Mythos Preview 17 hIndependentMETR Horizon v1.1 run 'gpt_4_turbo_inspect'; 95% CI 2.0–6.7 min

Compare every model on the same tests →

Also on these charts

Sources