AI models from OpenAI and Google DeepMind independently achieved gold medal scores at the 2025 International Math Olympiad (IMO). Both companies entered "informal" systems that ingested questions and generated proof-based answers in natural language.
OpenAI and Google researchers claimed these gold medal performances represent breakthroughs in AI reasoning models in non-verifiable domains. Google raised questions about OpenAI's announcement, citing a lack of official evaluation by IMO.
OpenAI used third-party evaluators to grade its AI model, while Google worked with IMO organizers and waited for official grading before announcing its results. The debate highlights how quickly AI models are improving, with both OpenAI and Google's AI models scoring higher than most high school students at IMO.