GPQA Diamond
Benchmark
Capacità
Google-proof PhD science questions — the reasoning era's exam.
Stai visualizzando un record di benchmark: un test usato dal settore per misurare chip o modelli — e cosa rivela concretamente.
GPQA's expert-written chemistry, physics and biology questions defeat search (PhDs outside the field score ~34%). Frontier reasoning models went from ~50% (2024) past 90% (late 2025) — a two-year sprint through a benchmark built to be hard.
Fonti
Record revisionato al 2026-01-15 — vedi Metodologia.