Benchmark
AMC 2022–23
Test mathematical reasoning on the AMC contest selection documented in the Gemini 1.5 evaluation.
Explore the original evidence
Explore the original evaluation, its methods, and reported findings.
Open evalCoverage
Google’s report specifies 250 competition problems: 100 AMC 12, 100 AMC 10 and 50 AMC 8. The original competitions are organized by the Mathematical Association of America.
Methodology
Gemini 1.5 reports a four-shot multiple-choice evaluation; section 6 lists the selection and appendix 12.13.5 documents the response protocol. Google is the evaluator; MAA is the competition organizer.