Google DeepMind’s “Gemini 2.5” Wins Gold at ICPC — Historic AI Programming Feat

DeepMind’s Gemini 2.5 model solved a complex programming problem at the International Collegiate Programming Contest (ICPC) that defeated all human competitors, marking a milestone in AI’s ability to do abstract reasoning and problem solving. While it didn’t clear all tasks, the win is being compared to Deep Blue’s chess victory and AlphaGo’s Go win. The Guardian