Google's AI, Gemini 2.5 Deep Think, just achieved a gold-medal level performance at the International Collegiate Programming Contest (ICPC)—basically the Olympics for college coders.
Competing remotely against 139 teams, Gemini finished in the top four by solving 10 out of 12 tough real-world coding problems in 677 minutes.
It even cracked Problem C—a challenge no human team managed to solve—by cleverly optimizing how liquids move through a network using dynamic programming and advanced search techniques.
Gemini's coding conquest
Gemini's performance highlights how far Google has come with training AI to think through complex problems step by step and double-check its work.
This isn't just about winning contests—it's another leap toward artificial general intelligence (AGI), where AI can handle any task humans can, not just follow scripts or patterns.