OpenAI’s Mysterious AI Wins Gold at IOI 2025
On August 12, tech outlet OfficeChai reported that OpenAI’s mysterious AI model delivered an outstanding performance at the 2025 International Olympiad in Informatics (IOI), clinching a gold medal. It ranked first among all AI participants and sixth overall on the global leaderboard.
For context, the IOI is one of the world’s most prestigious programming competitions for high school students. Over two intense days, participants must solve complex algorithm challenges while strictly adhering to time and memory limits — making the competition fiercely competitive.
OpenAI researcher Sheryl Hsu explained that their entry wasn’t specially trained for the IOI. Instead, it competed as part of a “cluster” of general-purpose reasoning models. The team only provided technical help for selecting which solutions to submit and connecting to the IOI API. Apart from that — including the way and timing of submissions — the entire process was run exactly the same as it would be for a human competitor.
At last year’s IOI, OpenAI’s model only placed in the 49th percentile — nowhere near the top three. This year, though, it scored 98 points, outperforming 98% of competitors and securing a gold medal. The same model has also won gold at the International Mathematical Olympiad (IMO) and taken second place in the AtCoder programming contest.