In a significant milestone, AI systems created by OpenAI and DeepMind achieved gold medal scores in the 2025 International Math Olympiad (IMO), a renowned competition typically dominated by the brightest young mathematicians globally.
This accomplishment highlights the swift advancement of AI capabilities, particularly in complex reasoning and problem-solving, which may have extensive implications across various sectors, including data security and market analysis.
This year, both organizations introduced "informal" systems that allowed their AI models to directly process questions and formulate proof-based responses in natural language, closely mimicking human comprehension and problem-solving abilities.
This innovation removed the necessity for human-machine translation, enabling the AI to independently analyze and resolve problems.
Both companies assert that their AI models outperformed the majority of high school students and Google’s previous-year AI, showcasing the rapid development of AI reasoning abilities.
Also Read: Japan, Australia, US Leaders Envision Powerful Defense Research Goals
Researchers participating in these IMO initiatives noted that these gold medal achievements signify important advancements in AI reasoning models, especially in non-verifiable domains.
While AI excels at tasks with definitive, verifiable solutions like mathematics or coding, these new systems are demonstrating potential in addressing more ambiguous challenges, although achieving true expertise in such tasks is still a goal for the future.
In addition to the remarkable mathematical successes, the IMO results have sparked another front in the ongoing AI competition: the struggle for public perception and leadership in the industry. Companies are aggressively vying to present themselves as frontrunners, an intangible yet vital battle for reputation that can greatly influence their capacity to attract top AI talent. Many leading AI researchers have roots in competitive mathematics, making benchmarks like the IMO particularly impactful.
Nonetheless, the announcements were not free of controversy. Google DeepMind raised concerns about the manner in which OpenAI announced its gold medal achievement, resulting in a public dispute reminiscent of adolescent rivalries. Demis Hassabis, CEO of Google DeepMind, along with other researchers, criticized OpenAI for revealing its accomplishment prematurely, shortly after the IMO disclosed student winners on Friday night and prior to official validation. Thang Luong, a senior researcher at Google DeepMind and the lead for their IMO project, pointed out that Google DeepMind delayed its announcement out of respect for the students and in accordance with the IMO Board’s request for official verification. Google collaborated with IMO organizers for over a year, seeking the IMO president’s approval and official grading. OpenAI, on the other hand, claimed they opted not to participate in a formal IMO competition earlier but enlisted three former IMO medalists as third-party evaluators for their model's performance.
Also Read: Behind the Lines of Indonesia's First High-Speed Railway System
The extraordinary gold medal achievements by AI models from OpenAI and Google DeepMind at the International Math Olympiad signify a crucial turning point in AI development. Despite the competitive tensions surrounding their announcements, the fundamental accomplishment is indisputable: AI is swiftly progressing in complex reasoning, transitioning from theoretical potential to actual reality. This advancement not only alters our understanding of machine intelligence but also suggests a future in which AI could transform every industry, from scientific exploration to day-to-day problem-solving.