Verdict
GPT-5.2 leads 18–2 across 20 shared benchmarks.
Claude Sonnet 4.5 2 · GPT-5.2 18 · higher isn't always better — check the caveats below.
Reported scores — protocols may differ. How we compare
Reasoning: GPT-5.2 leads 4–2 across 6 shared reasoning benchmarks (largest gap: ARC-AGI, 86.2 vs 63.67 — protocols undisclosed).
Math: GPT-5.2 leads 2–0 across 2 shared math benchmarks (largest gap: FrontierMath-Tier-4-v2-Private, 31.7 vs 2.44 — single evaluator).
GPT-5.2 leads 18–2 across 20 shared benchmarks. Claude Sonnet 4.5 leads 2 benchmarks and GPT-5.2 leads 18. Higher isn't always better — see the integrity caveats on each benchmark.
Claude Sonnet 4.5 and GPT-5.2 have 20 benchmarks in common in our data — those are the only rows where a direct, apples-to-apples comparison is drawn.
GPT-5.2 leads reasoning 4–2. GPT-5.2 leads math 2–0.