Atlas

Benchmarks

← All benchmarks

IMO 2025

Math · 2025-07-17

This MathArena track evaluates models on the IMO 2025 problem set. Scores report the percentage of problems answered correctly.

Top models (higher is better)

ModelScore
Gemini 2.5 Deep Think60.7
GPT-538.1
Gemini 2.5 Pro31.6
Grok 421.4
o316.7
O4 Mini14.3
DeepSeek-R1-05286.8
Loading Atlas data…