Google DeepMind’s Gemini 2.5 AI wins gold at international programming contest

Google DeepMind has achieved what it calls a “historic” AI breakthrough after its Gemini 2.5 model became the first AI to win a gold medal at an international programming competition, solving complex problems that stumped human programmers from top universities. The achievement represents a significant leap toward artificial general intelligence, with the model demonstrating advanced reasoning capabilities that could transform scientific and engineering disciplines.

What happened: The AI model competed against 139 of the world’s strongest college-level programmers at a competition in Azerbaijan, finishing second overall despite failing two of 12 tasks.

• In under 30 minutes, it solved a complex problem involving distributing liquid through interconnected reservoirs as quickly as possible—a challenge that required weighing infinite possibilities.
• Human teams from prestigious universities in Russia, China, and Japan failed to solve this particular problem.
• The model was specially trained for coding, mathematics, and reasoning problems, performing “as well as a top 20 coder in the world,” according to Google.

Why this matters: Google DeepMind executives compare this breakthrough to historic AI milestones like Deep Blue defeating chess champion Garry Kasparov in 1997 and AlphaGo beating a Go master in 2016.

• Unlike previous AI victories in constrained game environments, this achievement demonstrates reasoning applicable to real-world scenarios.
• The breakthrough could accelerate progress in drug discovery and chip design, according to the company.
• It marks a significant step toward artificial general intelligence—widely considered human-level intelligence across diverse tasks.

What they’re saying: Google DeepMind’s vice-president Quoc Le emphasized the real-world implications of the achievement.

• “For me it’s a moment that is equivalent to Deep Blue for Chess and AlphaGo for Go,” Le said. “Even bigger it is reasoning more towards the real world, not just a constrained environment.”
• “Solving complex tasks at these competitions requires deep abstract reasoning, creativity, the ability to synthesise novel solutions to problems never seen before and a genuine spark of ingenuity,” the company stated.
• Dr. Bill Poucher, executive director of the International Collegiate Programming Contest (ICPC), called it “a key moment in defining the AI tools and academic standards needed for the next generation.”

Expert skepticism: Leading AI researchers offered measured responses to Google’s breakthrough claims.

• Stuart Russell, professor of computer science at UC Berkeley, cautioned that “claims of epochal significance seem overblown,” noting AI systems have performed well on programming tasks before.
• However, Russell acknowledged the achievement may show “progress towards making AI-based coding systems sufficiently accurate for producing high-quality code.”
• Michael Wooldridge, professor at Oxford University, called it “an impressive achievement” but questioned the computational resources required.

The computational cost: Google declined to specify the computing power needed for the breakthrough, revealing only that it exceeded what’s available to subscribers of its $250-per-month Google AI Ultra service.

Historical context: The achievement follows four major AI breakthroughs, from the 1957 Perceptron through Deep Blue’s chess victory, AlphaGo’s Go triumph, and AlphaFold’s protein folding predictions that earned a Nobel Prize in 2024—positioning this programming achievement as the next milestone in AI’s evolution.