Skip to content

OpenAI

o4-mini

Announced
Access
Public

Benchmark scores

TestScoreBest on the trackerHow it was measured
Humanity's Last Examexpert questions, no tools18%GPT-6 Astra 55%Independentleaderboard entry "o4-mini (high) (April 2025)" (o4-mini-2025-04-16_high)
FrontierMath T1–3hard maths problems36%Claude Fable 5.2 95%Independenthigh reasoning effort (o4-mini-2025-04-16_high)Run 11 Jun 2026
GPQA DiamondPhD-level science questionsPhD experts: 70%Saturated80%Gemini 4 Argon 96%Independenthigh reasoning effort (o4-mini-2025-04-16_high)Run 16 Apr 2025
FrontierMath T4research-level maths problemsSaturated4.9%GPT-6.1 Sol 100%Independenthigh reasoning effort (o4-mini-2025-04-16_high)Run 11 Jun 2026
ARC-AGI-2visual puzzles, model aloneAverage person: 60%Saturated6.1%GPT-6 Astra 95%IndependentARC Prize config "o4-mini (High)" (o4-mini-2025-04-16-high)
OTIS Mock AIMEcompetition mathsSaturated82%GPT-5.5 100%Independenthigh reasoning effort (o4-mini-2025-04-16_high)Run 16 Apr 2025

Compare every model on the same tests →

Also on these charts

Sources