USAMO 2026

AI benchmark results and model performance

Type: percentageMath

88.10%

79.50% - 96.70%

79.50% - Claude Sonnet 5

Scores Over Time

Individual benchmark scores plotted by date.

Organisation	Model	Reported	Top Score	Info	Self Reported	Source
Anthropic	Claude Sonnet 5	30 Jun 2026	79.50%	High effort; 300k token limit; average over ten attempts per problem	Yes	Source
Anthropic	Claude Opus 4.8	28 May 2026	96.70%	Max effort	Yes	Source