OpenAI’s experimental large language model (LLM) has reached a remarkable achievement by earning gold medal-level performance at the 2025 International Math Olympiad (IMO). This milestone, shared by researcher Alexander Wei on X, underscores the advancements in AI reasoning capabilities. During the competition, the model tackled six challenging problems under strict conditions akin to human participants, solving five and securing 35 out of 42 possible points—graded by three former IMO medalists. Wei emphasized the model’s ability to handle high-level creative reasoning, noting the evolution of reasoning benchmarks leading to this success. While this experimental model won’t be publicly released yet, OpenAI is preparing to launch GPT-5 soon, which will feature a multi-model system for optimized task performance. Additionally, GPT-6 is already in training. This breakthrough demonstrates significant progress in AI mathematics, surpassing prior expectations and indicating the rapid advancement in technology.
Source link
