Mumbai, July 21: OpenAI's new experimental reasoning model has won the world's most prestigious math competition. The OpenAI LLM has won gold medal-level performance at the 2025 International Math Olympiad (IMO). OpenAI President Greg Brockman and Alexander Wei shared this remarkable achievement on social media, highlighting that the model took the same tests as humans to win the competition.

OpenAI CEO Sam Altman also confirmed, "We achieved gold medal level performance on the 2025 IMO competition with a general-purpose reasoning system!" He said that the LLM did the math and not a specific formal system. Altman said it was part of OpenAI's main push towards general intelligence. With this announcement, Sam Altman also confirmed launching the GPT-5 soon, setting accurate expectations. He said the experimental model incorporated new research techniques that were to be used in future models. Microsoft SharePoint Vulnerability: Tech Giant Issues Urgent Security Patch After Observing ‘Active Attacks’ on Document-Sharing Software Used by Government Agencies.

Sam Altman Announces OpenAI LLM Wins Gold Medal Level Performance at 2025 IMO Competition

OpenAI Experimental Reasoning LLM Wins Grand IMO 2025 Challenge

Gold Medal-Level Performance on 2025 IMO From Our Experimental Reasoning LLM: OpenAI President Greg Brockman

OpenAI's Experimental LLM Winning Gold Medal-Level Performance at IMO 2025

OpenAI CEO Sam Altman said that the experimental model achieved the world's most prestigious award at the 2025 International Math Olympiad; however, the company would not release it any time soon. He said it would take many months to release it to the masses.

OpenAI's experimental model was valued at under 4.5 hours without accessing the internet or other tools. The model had to write a detailed proof based on the IMO problems. As per the information available, the experimental model solved five out of six problems effectively, earning him 35 points out of 42. Each solution was assessed independently by three former IMO medalists. They gave the final scores after reaching a unanimous decision.

The International Math Olympiad (IMO) problems are considered to be the world's most difficult, and they require extended periods of "creative reasoning." According to the OpenAI researcher Alexander Wei , the "OpenAI IMO 2025 Proofs" were shared on GitHub. ‘You’re Not Unwell’: ChatGPT Triggers Manic Episode in Autistic Man After Validating Flawed Faster-Than-Light Theory, Admits Failure.

OpenAI shared the IMO Gold LLM experimental model details; however, it also hinted that it was a separate project. The OpenAI GPT-5 will be released soon, vastly improving the previous model and offering better reasoning, complex problem solving, and other improvements. Hyperbolic Labs co-founder Yuchen Jin said that GPT-6 was already training and may launch in the future.

(The above story first appeared on LatestLY on Jul 21, 2025 11:42 AM IST. For more news and updates on politics, world, sports, entertainment and lifestyle, log on to our website latestly.com).