OpenAI The mysterious generalized inference model that surprisingly attacked the IMO The 5 Puzzles of 2025, and it's successfully taken the gold? This news has made Greg Brockman and other big names retweet it with excitement. In other words, OpenAI may have developed a disruptive reasoning technology, and said goodbye to CoT for good, and there's one more piece of news: GPT-5 is coming.
Recently, the world's top big models were still all over the 2025 IMO field, not even touching the bronze medal.
However, OpenAI dropped a bombshell - they managed to take the gold medal at IMO 2025 with a brand new 'Universal Reasoning Model'!
6 questions, 5 solved, 35 points!
Only six human participants could make this recognized problem. o3, o4-mini, Gemini 2.5 Pro, Grok-4, and DeepSeek-R1 also scored zero points for all six big models, and even OpenAI's super-powerful model similarly folded.
Even more shockingly, OpenAI announced that the model that made history is not the rumored GPT-5, but a brand new, experimental model!