Verdict
GPT-5 mini leads 12–5 across 17 shared benchmarks.
GPT-5 mini 12 · o4-mini 5 · higher isn't always better — check the caveats below.
Reported scores — protocols may differ. How we compare
Reasoning: Evenly split 2–2 across 4 shared reasoning benchmarks (largest gap: GPQA diamond, 72.81 vs 66.67 — single evaluator).
Math: GPT-5 mini leads 3–0 across 3 shared math benchmarks (largest gap: FrontierMath-Tier-4-v2-Private, 12.2 vs 4.88 — single evaluator).
Unknown: GPT-5 mini leads 3–0 across 3 shared unknown benchmarks (largest gap: LMCA, 40.24 vs 31.18 — single evaluator).
GPT-5 mini leads 12–5 across 17 shared benchmarks. GPT-5 mini leads 12 benchmarks and o4-mini leads 5. Higher isn't always better — see the integrity caveats on each benchmark.
GPT-5 mini and o4-mini have 17 benchmarks in common in our data — those are the only rows where a direct, apples-to-apples comparison is drawn.
GPT-5 mini leads math 3–0. GPT-5 mini leads unknown 3–0.