The breakthrough by OpenAI: AI wins an IMO gold medal

OpenAI announced that its internal experimental model performed exceptionally well at the 2025 International Mathematical Olympiad (IMO), achieving results comparable to human gold medalists by successfully solving 5 out of 6 problems, with a total score of 35 out of 42. This achievement demonstrates a breakthrough in AI’s ability to engage in prolonged creative thinking and tackle complex proof tasks, showing significant improvement compared to traditional reinforcement learning methods. It is worth noting that the upcoming GPT-5 has different capabilities in mathematical reasoning, despite high expectations surrounding it. Additionally, OpenAI researcher Alexander Wei pointed out that AI development is progressing faster than anticipated, especially in mathematical benchmark performance, indicating rapid advancement toward general artificial intelligence.