Solving Problems Human Geniuses Couldn’t in Just 30 Minutes? Google Gemini Wins ‘Gold Medal’ at the World’s Best Coding Competition

Imagine the smartest university students from around the world gathered in one place. These are ‘coding masters’ who have spent years sharpening their weapons of mathematics, logic, and programming. But what if a single computer sitting in the corner solved a problem in just 30 minutes—a challenge that these experts couldn’t crack even after five hours of intense collaboration?

This isn’t a scene from a science fiction movie. It actually happened at the ‘International Collegiate Programming Contest (ICPC)’ World Finals in 2025. The protagonist of this story is Google DeepMind’s latest artificial intelligence, Gemini 2.5 Deep Think.

Why Does This Matter?

The weight of this event is far greater than simply thinking, “AI is getting good at coding.” It is on a completely different level from the capabilities of the chatbots we have seen so far.

Usually, when we ask an AI to “write code to build this website,” the AI ‘assembles’ similar code based on the vast amount of data it has already learned. To use an analogy, it’s like a chef who cooks by memorizing existing recipes. However, the problems presented at the ICPC are different. This competition is the ultimate arena for algorithms (logical procedures for solving problems).

Solving a problem here doesn’t just mean typing fast; it requires ‘reasoning ability’—the capacity to draw one’s own map to find treasure in a complex maze and logically untangle knotted threads. [Source 3] Google CEO Sundar Pichai described this as a “profound leap.” [Source 11]

In other words, it means that AI has begun to stand shoulder-to-shoulder with the world’s greatest human geniuses in the realms of ‘deep thinking’ and ‘complex logical reasoning’—areas once believed to be uniquely human. [Source 3, Source 5] Put simply, it’s a signal that AI has moved beyond the stage of being good at ‘memorization’ and has truly begun to ‘think.’

Easy Understanding: What Happened at the ‘Olympics of Coding’

1. What is the ICPC?

The ICPC is the dream stage for university students worldwide. It is so prestigious that it’s called the ‘Olympics of Computer Programming,’ and winning an award there is enough to be recognized as one of the world’s top engineers. [Source 2, Source 6] The barrier to entry is extremely high, with only about 140 teams out of tens of thousands making it to the finals, and the 12 problems presented are known to test the very limits of human capability. [Source 10]

2. Gemini’s Report Card: 10 out of 12 Problems Solved

In this competition, Gemini correctly solved 10 out of 12 problems. [Source 7, Source 8] Consequently, it recorded a massive score corresponding to 2nd place overall when compared to the human participants, proving its ‘gold medal-level’ performance. [Source 8, Source 9] Achieving a gold medal-tier result in a field of the world’s brightest minds is a monumental record in the history of AI.

3. Solving the Problem Everyone Else Gave Up On

The most chilling moment was separate. There was one ultra-high-difficulty problem that not a single human team could solve. Gemini was the only one to crack it. [Source 9, Source 12] This demonstrates that AI can go beyond listing knowledge or finding patterns to find new logical breakthroughs that even humans haven’t reached. [Source 9]

4. Overwhelming Speed and a Handicap

Gemini also overwhelmed humans in terms of time management.

• 10-Minute Handicap: To ensure fairness, Gemini started 10 minutes later than the human teams. [Source 7]
• High-Speed Solutions: Despite this, it poured out 8 correct answers within just 45 minutes of starting. [Source 7]
• 30-Minute Miracle: In particular, one complex problem took less than 30 minutes to solve. [Source 1] It solved in the time it takes to drink a cup of coffee what humans would struggle with for hours.

Current State: The Emergence of ‘Agentic’ AI

The model deployed this time isn’t just a standard Gemini that answers questions. It is an advanced version of ‘Gemini 2.5 Deep Think’ that utilizes an agentic approach (a structure that sets its own goals, uses tools, and takes action). [Source 3, Source 4]

To use another analogy: if previous AI was like a ‘new intern’ who could barely do what they were told, this model is like a ‘veteran problem solver’ who determines what is needed, gathers tools, and persists until the problem is solved. When given a problem, it forms a hypothesis, writes code, analyzes why it failed if there’s an error, and repeats the process of correction. Vahab Mirrokni of Google DeepMind emphasized this as a “great achievement in complex algorithmic reasoning and coding ability.” [Source 3]

These changes show that AI is evolving from a passive tool into an active subject capable of defining and solving problems on its own. [Source 5]

What Happens Next?

We are now at a point where the relationship between AI and humans is being completely redefined. This achievement shouldn’t be viewed merely as a competition where “AI is better at coding than humans.” Experts believe this event will be a ‘key moment in defining AI tools and academic standards for the next generation.’ [Source 5]

Imagine this: the AI assistants we meet in the future won’t just perform simple errands like “set a schedule.” They will become ‘the best partners’ who deliberate with us on the complex and logical problems we face in business or research, suggesting brilliant breakthroughs that humans might never have considered.

The reason Google CEO Sundar Pichai called this achievement a “profound leap” while celebrating may be that AI has truly begun to learn ‘how to think’ like a human. [Source 11]

Perspective of MindTickleBytes’ AI Reporter

The news of the gold medal at the ICPC symbolizes AI’s evolution from an ‘encyclopedia of knowledge’ to a ‘solver of wisdom.’ The fact that AI alone solved a problem that all human teams abandoned gives us much to think about.

This serves as a powerful message of hope: AI may bring us unimaginable answers to the many challenges facing humanity today, such as climate change, incurable diseases, and energy issues. If human intuition and AI’s sophisticated logic are combined, we will be able to achieve the progress of civilization at a faster rate than ever before.

Of course, this is also the time to begin serious discussions on how humans can responsibly control and utilize such ‘super-intelligent’ tools. Great power, after all, comes with great responsibility.

References

1. Google Deepmind’s Gemini 2.5 AI wins gold medal at ICPC - Overview
2. Gemini wins International Collegiate Programming Contest gold
3. Gemini achieves gold-level performance at the International… (LinkedIn)
4. Gemini achieves gold-medal level at the International Collegiate Programming Contest World Finals
5. Gemini shows off at another high-level academic competition
6. Google Gemini Achieves Gold-Medal Performance at International…
7. Gemini achieves gold-medal level at the Programming Contest…
8. Gemini just aced the world’s most elite coding competition - ZDNET
9. Gemini AI Achieves Gold Medal Performance at Prestigious Programming …
10. Gemini 2.5 Deep Think achieves gold-medal level at ICPC 2025
11. Google CEO Sundar Pichai celebrates Gemini’s gold win at world coding …
12. Gemini 2.5 Achieves Gold Medal Performance at ICPC World Finals

FACT-CHECK SUMMARY

• Claims checked: 18
• Claims verified: 18
• Verdict: PASS