Experts Mission

GPQA Diamond

PhD-level scientific reasoning.

Get in touch Read research

198Public tasks

94.2%Highest score

The GPQA Diamond leaderboard

448 graduate-level, 'Google-proof' multiple-choice science questions written by domain experts.

Paper Data Code Sample task

Model

Score

Gemini 3.1 Pro

Gemini 3.1 ProHigh

94.2%

Gemini 3.6 Flash

Gemini 3.6 FlashHigh

93.4%

Kimi K3

Kimi K3Max

92.7%

Grok 4.5

Grok 4.5High

92.4%

Opus 4.8

Opus 4.8Max

92.0%

70%

80%

90%

100%