Atlas

Benchmarks

← All benchmarks

AIME 2024 (with tools)

Math · 2024-02-07

AIME 2024's 30 problems from the 2024 American Invitational Mathematics Examination solved with tools available, usually a Python interpreter or code execution; some reports do not itemize the tools. Tool runs approach a perfect score and are not comparable with runs without tools, which are scored under AIME 2024.

Benchmark source: Maa

Top models (higher is better)

ModelScoreSource
o4-mini98.7gpt-oss launch post · OpenAI
gpt-oss-120b96.6gpt-oss launch post · OpenAI
gpt-oss-20b96.0gpt-oss launch post · OpenAI
o395.2gpt-oss launch post · OpenAI
Loading Atlas data…