Gemini achieves gold-medal level at the ICPC World Finals

Gemini 2.5 Deep Think achieves breakthrough performance at the world’s most prestigious computer programming competition, demonstrating a profound leap in abstract problem solving.

An advanced version of Gemini 2.5 Deep Think has achieved gold-medal level performance at the 2025 International Collegiate Programming Contest (ICPC) World Finals.

This milestone builds directly on Gemini 2.5 Deep Think's gold-medal win at the International Mathematical Olympiad (IMO) just two months ago. Innovations from these efforts will continue to be integrated into future versions of Gemini Deep Think, expanding the frontier of advanced AI capabilities accessible to students and researchers.

Solving complex tasks at these competitions requires deep abstract reasoning, creativity, the ability to synthesize novel solutions to problems never seen before and a genuine spark of ingenuity.

Together, these breakthroughs in competitive programming and mathematical reasoning demonstrate Gemini’s profound leap in abstract problem-solving — marking a significant step on our path toward artificial general intelligence (AGI).

ICPC sets a global standard for excellence

The ICPC is globally recognized as the oldest, largest and most prestigious algorithmic programming competition at college level. This is a step up from high school level olympiads such as the IMO. Every year, participants from nearly 3000 universities and over 103 countries compete in solving real-world coding problems.

This year’s world finals took place in Baku, Azerbaijan on September 4, and brought together the top teams from earlier phases of the competition. Over a five-hour period, each team tackled a set of complex algorithmic problems. Final rankings hinged on two unforgiving principles: only perfect solutions earned points, and every minute counted. From the 139 competing teams, only the top four teams won gold medals.

Gemini solved 10 of 12 problems, achieving gold-medal level

An advanced version of Gemini 2.5 Deep Think competed live in a remote online environment following ICPC rules, under the guidance of the competition organizers. It started 10 minutes after the human contestants and correctly solved 10 out of 12 problems, achieving gold-medal level performance under the same five-hour time constraint. See our solutions here.

Gemini solved eight problems within just 45 minutes and two more problems within three hours, using a wide variety of advanced data structures and algorithms to generate its solutions. By solving 10 problems in a combined total time of 677 minutes, Gemini 2.5 Deep Think would be ranked in 2nd place overall, if compared with the university teams in the competition.

Dr. Bill Poucher, ICPC Global Executive Director, stated: “The ICPC has always been about setting the highest standards in problem solving. Gemini successfully joining this arena, and achieving gold-level results, marks a key moment in defining the AI tools and academic standards needed for the next generation. Congratulations to Google DeepMind; this work will help us fuel a digital renaissance for the benefit of all.”

Gemini solved a problem no university team solved

In an unprecedented moment, our model successfully and efficiently solved Problem C within the first half hour — which no university teams in the contest solved.

Problem C required finding a solution for distributing liquid through a network of interconnected ducts to a set of reservoirs, with the goal of finding a configuration of these ducts that fills all the reservoirs as quickly as possible. There are an infinite number of possible configurations, as each duct may be open, closed or even partially open, making it very difficult to search for the optimal configuration.

Gemini found an effective solution with a clever insight: it first assumed each reservoir has a "priority value" representing how much each reservoir should be favored compared to the others. When given a set of priority values, the best configuration of the ducts can be found using a dynamic programming algorithm. Gemini discovered that by applying the minimax theorem, the original problem can be approached by finding the priority values that make the resulting flow most constrained. Leveraging the relationship between priority values and optimal flows, Gemini used nested ternary searches to quickly find optimal priority values in the bowl-like convex solution space, and solved Problem C.

Gemini’s performance brings together a series of advances

Our milestone performance brings together a series of advances across pretraining, post-training, novel reinforcement learning techniques, multi-step reasoning and parallel thinking. These innovations helped Gemini explore different ways of solving complex problems, verifying solutions and continuously iterating before responding.

For example, during the course of reinforcement learning, we trained Gemini to reason and generate code for some of the most difficult problems coders have faced, to learn from feedback on results and evolve its approaches. To tackle a problem, multiple Gemini agents each propose their own solutions, use terminals to execute code and tests, and then iterate the solutions based on all attempts.

Our internal studies show that a similar version of Gemini 2.5 Deep Think can also achieve gold-medal level performance in the 2023 and 2024 ICPC World Finals, performing as well as the world's top 20 competitive coders.

Exploring Gemini's potential as a collaborator

Achieving gold-medal level at the ICPC has immediate, practical consequences for software development and shows that AI can act as a true problem-solving partner for programmers. If the best AI and human solutions in the competition were combined, all 12 problems would have been solved completely and correctly. This shows the potential for AI to provide unique, novel contributions that complement the skills and knowledge of human experts.

Beyond math and coding, our achievement demonstrates a powerful new capability in abstract reasoning. The skills needed for the ICPC — understanding a complex problem, devising a multi-step logical plan and implementing it flawlessly — are the same skills needed in many scientific and engineering fields, such as designing new drugs or microchips. It shows that AI is moving from just processing information to actually helping solve some of the world's most difficult reasoning problems in ways that could benefit humanity.

Gemini users with Google AI Ultra subscriptions can already use a lightweight version of Gemini 2.5 Deep Think in the Gemini app. And in the near future, much smarter AI coding assistants could help developers tackle increasingly complex engineering challenges. From logistics and debugging to scientific research, solutions to some of the hardest, most unsolvable problems may soon be within reach by using AI as a collaborative tool.