AI Math Showdown: OpenAI and Google Both Earn Gold, Neither Laps the Other

Share
AI Math Showdown: OpenAI and Google Both Earn Gold, Neither Laps the Other

In a remarkable display of artificial intelligence, both OpenAI and Google DeepMind have clinched gold-medal scores on the 2025 International Math Olympiad (IMO) — one of the most demanding high school math contests in the world.

Breaking Through Complex Problem-Solving

This year's IMO challenge was tackled by “informal” AI models capable of reading problems in natural language and producing step-by-step proofs — no specialized coding or formatting required. Impressively, both companies' systems correctly solved five out of six problems, outperforming most human competitors and even besting Google's previous formal system that won silver last year.

A Signal of AI Reasoning Maturity

The achievement highlights how advanced AI reasoning skills have become. While AI has excelled at structured tasks like straightforward computation and coding, handling ambiguous, proof-based problems without a strict answer key marks a new level of sophistication.

A Clash Over Timing and Transparency

Despite shared success, tensions arose between the two AI giants over the announcement process. Google's DeepMind deliberately waited until official IMO grading was confirmed, aligning releases to avoid overshadowing students — a decision CEO Demis Hassabis emphasized publicly.

OpenAI, by contrast, released its gold-medal results earlier after having independent former medalists validate its performance. According to OpenAI, the IMO organizers asked the company to wait until competitors received their awards — a request they honored before their announcement.

This friendly rivalry demonstrates that both OpenAI and Google are neck-and-neck in conquering complex reasoning benchmarks. With AI models rapidly closing gaps in sophisticated domains like informal proofs, the race for next-generation AI capabilities is clearly still wide open.

Read more