List your company

STEM Reasoning on GPQA-Diamond (Pass@1)

91.9Pass@1Gemini 3-pro
28.56445.00761.4577.893Mar 21, 2026

Evaluation results

MethodLinks
Gemini 3-proDataset=GPQA-Diamond2026.0391.9
Gemini 2.5-proDataset=GPQA-Diamond2026.0386.4
GPT-5 HighDataset=GPQA-Diamond2026.0385.7
Seed1.8Dataset=GPQA-Diamond2026.0383.8
Claude Sonnet-4.5Dataset=GPQA-Diamond2026.0383.4
Gemini 3-proDataset=PHYBench2026.0359
Gemini 3-proDataset=BIOBench2026.0351.9
Gemini 2.5-proDataset=PHYBench2026.0348
GPT-5 HighDataset=BIOBench2026.0348
Claude Sonnet-4.5Dataset=BIOBench2026.0344.6
Seed1.8Dataset=BIOBench2026.0342.3
Gemini 2.5-proDataset=BIOBench2026.0341.5
Seed1.8Dataset=PHYBench2026.0341
GPT-5 HighDataset=PHYBench2026.0340
Claude Sonnet-4.5Dataset=PHYBench2026.0331