Next Article
Humans outsmart AI at International Mathematical Olympiad, score perfect 42/42
Five teens just pulled off something wild at the 2025 International Mathematical Olympiad in Australia—they scored a perfect 42/42, outsmarting even top AI models like Google's Gemini and OpenAI's ChatGPT.
No AI has ever managed a perfect score here.
How did the humans and AI fare?
The IMO is no joke: 641 students from 112 countries had to solve six tough math problems in under five hours.
While Gemini and ChatGPT did well (35/42, which still gets gold), they couldn't quite match human perfection—though Gemini did way better than last year.
The competition made sure grading was totally fair between humans and AI, so these wins really show what young minds can do under pressure.