OpenAI vs Google: AI Rivals Excel in Math but Still Tied
▼ Summary
– OpenAI and Google DeepMind’s AI models achieved gold-medal scores in the 2025 International Math Olympiad (IMO), showcasing rapid advancements in AI reasoning.
– Both companies used “informal” systems this year, allowing their AI models to process questions and generate answers in natural language without human translation.
– Google criticized OpenAI for announcing its results prematurely and without official IMO evaluation, while Google followed a more rigorous process with official grading.
– Researchers highlighted that the AI models’ performance represents a breakthrough in reasoning for non-verifiable domains, though they still struggle with ambiguous tasks.
– The competition underscores how closely matched OpenAI and Google are in the AI race, with both aiming to secure top talent and industry leadership.
The competition between OpenAI and Google DeepMind has reached new heights, with both AI giants achieving gold-medal performances at the prestigious International Math Olympiad (IMO). This milestone highlights the rapid progress in artificial intelligence, particularly in complex reasoning tasks that were once considered beyond the reach of machines. The 2025 IMO results reveal not just technological breakthroughs but also the intense rivalry shaping the future of AI development.
Both companies entered AI systems capable of solving high-level math problems without human intervention, a significant leap from previous years. OpenAI and Google’s models each correctly answered five out of six competition questions, outperforming most human participants and demonstrating advanced problem-solving abilities. Unlike last year, when Google’s system required manual translation of problems into machine-readable formats, this year’s models processed questions in natural language and generated proofs independently.
The achievement underscores AI’s growing proficiency in formal reasoning, though challenges remain in less structured domains. While these systems excel at precise mathematical tasks, they still struggle with open-ended problems requiring real-world judgment. The IMO success serves as a key benchmark for AI capabilities, especially in recruiting top talent, many leading researchers have backgrounds in competitive mathematics.
However, the celebration wasn’t without controversy. Google criticized OpenAI for announcing its results prematurely, before the IMO’s official verification process concluded. Google’s team emphasized their adherence to competition protocols, waiting for independent validation before publicizing their model’s performance. OpenAI defended its approach, stating it relied on evaluations from former IMO medalists and coordinated with organizers before sharing its findings.
The debate over methodology aside, the broader takeaway is clear: AI is advancing at an unprecedented pace, with multiple labs now producing systems capable of elite-level reasoning. As OpenAI prepares to launch GPT-5 and Google continues refining its models, the race for AI supremacy remains fiercely contested. What was once a clear lead for OpenAI has evolved into a neck-and-neck competition, proving that innovation in artificial intelligence is far from a one-company game.
(Source: TechCrunch)
