Google's Gemini Deep Think Achieves Math Competition Gold Medal with Advanced Version

Google's Gemini Deep Think model has achieved a gold-medal level performance in the International Mathematical Olympiad (IMO) competition. The model, which uses advanced reasoning techniques and can operate end-to-end in natural language, solved five out of six problems perfectly and earned 35 total points. A version of the model will be made available to trusted testers and will eventually be released to subscribers of the Google AI Ultra service.

Google's Gemini Deep Think model has made a significant stride in the realm of artificial intelligence by achieving a gold-medal level performance in the International Mathematical Olympiad (IMO) competition. The model, developed by Google DeepMind, demonstrated exceptional problem-solving capabilities by solving five out of six problems perfectly, earning a total of 35 points. This performance is particularly notable as it showcases the model's ability to operate end-to-end in natural language, a critical advancement in AI technology [1].

The IMO, an elite math competition for high school students, draws participants from over 100 countries. The competition's rigorous nature and high standards make the achievement of Google's model even more impressive. The model's success indicates the remarkable progress of general-purpose AI models in solving advanced mathematical problems [1].

Google DeepMind's senior staff research scientist, Thang Luong, stated that the model used in the IMO is "actually very close to the main Gemini model that we have been offering to people." This suggests that the model's performance is a strong indicator of the potential of Google's AI technology [1].

The success of the Gemini Deep Think model in the IMO competition highlights the growing capabilities of AI in solving complex problems. While the model's performance is experimental and not yet available to the public, Google plans to make it available to trusted testers soon, with a broader release expected for subscribers of the Google AI Ultra service [1].

The IMO competition is not just about showcasing AI's capabilities but also about promoting the beauty of mathematics to high school students and encouraging them to pursue careers in the field. Google DeepMind's participation and success in the competition reflect its commitment to advancing both AI technology and education [1].

The financial implications of this technological advancement are significant. As AI continues to evolve, its applications in various industries, including finance, healthcare, and education, will expand. Companies like Google, which are at the forefront of AI development, stand to benefit from the increasing demand for AI-driven solutions [1].

In conclusion, Google's Gemini Deep Think model's gold-medal performance in the IMO competition is a testament to the remarkable progress of AI technology. The model's ability to solve complex mathematical problems in natural language is a significant milestone that could pave the way for future innovations in AI. As the technology continues to advance, its impact on various industries and the global economy is likely to be substantial.

References:

[1] https://www.axios.com/2025/07/21/openai-deepmind-math-olympiad-ai