In what could be considered as a landmark achievement for AI, both Google and OpenAI have achieved gold-medal level performances at prestigious international programming competitions, which is also known more popularly as the ‘coding Olympics.’
Google DeepMind’s advanced Gemini 2.5 Deep Think model secured gold at the 2025 International Collegiate Programming Contest (ICPC) World Finals. On the other hand, an OpenAI system earned gold at the International Olympiad in Informatics (IOI). The impressive performance has shown how advanced these AI models are compared to the rest of the world.
Google’s Gemini wins ICPC 2025
The ICPC, which was held this year in Baku, Azerbaijan, gathers the world’s top university programming teams to solve complex algorithmic challenges within a five-hour timed competition. Google’s Gemini 2.5 Deep Think impressed everyone by solving 10 out of 12 difficult problems, including a highly intricate optimisation problem that no human team cracked, thus demonstrating the superior capacity for abstract reasoning and coding under pressure.
It’s worth noting that the model started 10 minutes after humans but rapidly outperformed most teams, showcasing AI’s rising dominance in competitive programming.
In August 2025, OpenAI’s system won gold at the IOI – the top programming contest for high school students worldwide. Their AI matched the performance of elite human competitors, solving all 12 IOI problems. This was a leap from near misses in previous years.
ICPC not the only win for Gemini
The ICPC achievements follow DeepMind’s July 2025 gold-medal performance at the International Mathematical Olympiad (IMO) where Gemini, with an enhanced Deep Think architecture, solved five out of six challenging math problems at a level comparable to top human contestants.
Unlike earlier tools that required translating problems into code-like formats, these models work end-to-end through natural language, generating multi-step logical reasoning and algorithmic code independently.
