OpenAI Showed IMO Gold Level Math Skills Through Its LLM
OpenAI has developed an experimental large language model (LLM) that showed gold medal level performance in the 2025 International Math Olympiad (IMO). This model scored 35 out of 42 marks, which is considered enough for a gold medal in the actual competition.
- OpenAI Showed IMO Gold Level Math Skills Through Its LLM
- OpenAI Model Explained Complex Proofs in Natural Language
- OpenAI Used a New Training Method to Solve Math Problems
- Sam Altman Confirmed GPT 5 Launch but IMO Level Model Will Not Be Released Now
- OpenAI LLM Has Set a New Benchmark by Reaching IMO Gold Level
- OpenAI LLM Can Now Solve Real World Math Challenges
1/N I’m excited to share that our latest @OpenAI experimental reasoning LLM has achieved a longstanding grand challenge in AI: gold medal-level performance on the world’s most prestigious math competition—the International Math Olympiad (IMO). pic.twitter.com/SG3k6EknaC
— Alexander Wei (@alexwei_) July 19, 2025
OpenAI researcher Alexander Wei said that they tested the model under the same exam conditions used for humans. They used official IMO questions and conducted two exam sessions of 4.5 hours each. The model solved problems without using any tool, internet or external help.
OpenAI Model Explained Complex Proofs in Natural Language
The model solved the first five out of six IMO problems. It could not solve the last problem. The model wrote all answers in natural language with clear steps.
Three former IMO gold medallists checked the model’s answers independently. All three said the model’s solutions were correct. They confirmed that the model used proper reasoning and step-by-step explanation.
OpenAI Used a New Training Method to Solve Math Problems
OpenAI said that this model was not trained using traditional reinforcement learning. The company trained the model to write complex and long proofs by itself.
Alexander Wei said that IMO problems cannot be solved with only calculations. They need deep creative reasoning. The model showed that AI can now write logical and structured arguments like human mathematicians.
Sam Altman Confirmed GPT 5 Launch but IMO Level Model Will Not Be Released Now
we achieved gold medal level performance on the 2025 IMO competition with a general-purpose reasoning system! to emphasize, this is an LLM doing math and not a specific formal math system; it is part of our main push towards general intelligence.
when we first started openai,… https://t.co/X46rspI4l6
— Sam Altman (@sama) July 19, 2025
OpenAI CEO Sam Altman said that GPT 5 will be launched soon. But he also made it clear that the model which reached IMO level will not be released to the public right now.
Sam said this model is part of a separate research track. The company has no current plan to release it. GPT 5 will be a general purpose reasoning model, but it will not include IMO level performance yet.
OpenAI LLM Has Set a New Benchmark by Reaching IMO Gold Level
Alexander Wei shared that AI progress was much faster than expected. In 2021, his PhD advisor asked him to predict AI performance in math. He expected only 30 percent score on the MATH benchmark by 2025. But now the model has reached IMO gold level.
Alexander also thanked his team members Sheryl Hsu and Noam Brown for their help. He congratulated all IMO 2025 participants and said that many OpenAI researchers were also former IMO medallists.
OpenAI LLM Can Now Solve Real World Math Challenges
OpenAI has shown that AI can now perform at the top level in exams that were earlier dominated only by humans. GPT 5 is just the beginning but this experimental model has proved that AI can now match humans in creative reasoning.
This milestone by OpenAI is not just important for AI progress but also for the future of education, research and logic based learning.
ALSO READ: Nothing Phone 3 vs OnePlus 13: Best Flagship of 2025
