Google’s Gemini and OpenAI’s ChatGPT have achieved gold medal-level scores in the prestigious International Math Olympiad (IMO) 2025. Both AI models successfully solved five out of six highly complex mathematical problems, scoring 35 out of 42 marks — a score considered sufficient to earn a gold medal.
AI Excels in Complex Math Problems
Google revealed that its Gemini Deep Think model participated in the Olympiad following all the official rules, without any human assistance. The AI completed the tasks within the 4.5-hour time limit, producing mathematical proofs directly from problem descriptions using natural language. OpenAI, on the other hand, used an experimental research model that also delivered similar results. The scores were evaluated independently by a panel of former IMO medalists and finalized with unanimous consensus.
About the International Math Olympiad
The International Math Olympiad, first held in 1959 in Romania, is one of the oldest and most respected mathematics competitions for school students. More than 100 countries participate every year. The competition is unique because it focuses on mathematical proofs rather than direct solutions. Participants must use logic, advanced theorems, and applied mathematics to construct detailed proofs, which are then graded for accuracy and depth.
Future of AI in Mathematics
Google DeepMind CEO Demis Hassabis stated that Gemini’s success demonstrates the ability of AI to understand and solve high-level mathematical problems end-to-end. The enhanced Gemini Deep Think model will soon be available to select mathematicians and testers before its broader release to Google AI Ultra subscribers.
OpenAI also made headlines by unofficially taking part in the test. According to TechCrunch, the company hired three former IMO gold medalists as independent evaluators to ensure the accuracy of its results before submitting them to the IMO. Alexander Wei, a member of OpenAI’s technical staff, confirmed that the results were verified and agreed upon by all evaluators.
This achievement highlights a new era in AI capabilities, where advanced chatbots like Gemini and ChatGPT can compete at the highest levels of problem-solving and logical reasoning, pushing the boundaries of what artificial intelligence can achieve in academic and scientific fields.
