
Google AI’s Gold-Medal Performance

AI Breakthrough
Google's artificial intelligence system achieved gold-medal level performance at a major international computer programming competition.
Context
The International Collegiate Programming Contest is considered the world's most prestigious university-level programming competition. The contest has run for decades and draws teams from nearly 3,000 universities across more than 100 countries. Past participants have included Google's co-founder and other prominent figures in technology.
Gold Medal Performance
On Wednesday, the CEO and co-founder of Google DeepMind, Nobel laureate Demis Hassabis, announced that the AI model Gemini 2.5 Deep Think correctly solved 10 out of 12 problems during the five-hour competition held in Azerbaijan.
While the AI system was not an official competitor, its performance would have placed the AI in second place overall – behind Russia's St. Petersburg State University – if ranked against the 139 university teams that competed. Only the top four teams earned gold medals at the event.
Solving the Unsolvable
Gemini accomplished something no human team could do. Google's AI solved one problem – finding the best way to distribute liquid through a network of pipes to multiple containers as quickly as possible – within the first half hour of the competition. The AI used advanced mathematical techniques to find a solution that had eluded all human competitors.
Previous Success
This achievement came just two months after the Gemini 2.5 Deep Think model won gold at the International Mathematical Olympiad. Google trained the model by having it practice on extremely difficult problems and learn from its mistakes. The system used multiple AI agents working together to propose different solutions and test them before selecting the best approach. Google believes that these milestones show how AI can begin to tackle some of the world's most complex reasoning and "unsolvable" problems for the benefit of humanity.