OpenAI said its newest reasoning system achieved a gold-medal score at the 2025 International Olympiad in Informatics, placing sixth overall among about 330 human contestants and first among artificial-intelligence entries. The showing marks a sharp jump from last year, when an earlier version finished just below the bronze-medal cutoff at the 49th percentile. This year’s 98th-percentile result was reached without code or data tuned specifically for IOI problems, underscoring improvements in the model’s general reasoning abilities, the company said. The programming win extends a string of recent benchmark gains: the same system recently took second place on the AtCoder programming platform and delivered gold-medal-level performance on the International Mathematical Olympiad. OpenAI said the breadth of results suggests its models are beginning to generalize across technical domains.