OpenAI and Google AI Win Gold in Math Olympiad 2025

Key Points

• OpenAI and Google AI Win Gold in Math Olympiad 2025
• Both AI models solved 5 out of 6 IMO questions
• Google took a formal route; OpenAI moved faster
• Debate rises over grading and announcement timing
• AI now competes with top global math students

OpenAI and Google’s AI models have reached a new milestone — each earning gold-medal-level scores at the 2025 International Math Olympiad (IMO), one of the most prestigious high school math competitions globally.

These AI systems, developed independently by OpenAI and Google DeepMind, solved five out of six problems in this year’s IMO test.

1/N I’m excited to share that our latest @OpenAI experimental reasoning LLM has achieved a longstanding grand challenge in AI: gold medal-level performance on the world’s most prestigious math competition—the International Math Olympiad (IMO). pic.twitter.com/SG3k6EknaC

— Alexander Wei (@alexwei_) July 19, 2025

That performance places them ahead of most human participants — and equal with each other — in a contest known for its complexity and rigor.

What makes this even more impressive? This year, both companies used “informal” AI systems.

Unlike last year, when Google’s system needed humans to convert questions into machine-friendly formats, the 2025 models directly read natural language problems and produced natural language proofs. No translations. Just AI thinking like a top-tier mathlete.

Both OpenAI and Google describe this as a major step in AI reasoning — especially in “non-verifiable” domains where there’s no simple right answer.

Official results are in – Gemini achieved gold-medal level in the International Mathematical Olympiad! 🏆 An advanced version was able to solve 5 out of 6 problems. Incredible progress – huge congrats to @lmthang and the team! https://t.co/pp9bXF7rVj

— Demis Hassabis (@demishassabis) July 21, 2025

While AI has already shown strength in solving structured problems (like coding or reasoning tasks through ChatGPT agents), the IMO challenge pushes it toward more abstract and human-style reasoning.

Tensions Rise Over Credibility and Announcement Timing

While the scores may be even, the rollout of results wasn’t. And that’s where things get interesting.

OpenAI made headlines Saturday morning with its gold-medal announcement, just hours after IMO announced which high schoolers had won. But that raised eyebrows — especially at Google.

Google DeepMind’s CEO and researchers went public, criticizing OpenAI for jumping the gun. They argue that OpenAI didn’t follow the official IMO grading process and released its results without proper validation.

Google, on the other hand, says it worked closely with the competition’s organizers and waited until Monday — after getting approval and grading from IMO — before going public.

Btw as an aside, we didn’t announce on Friday because we respected the IMO Board’s original request that all AI labs share their results only after the official results had been verified by independent experts & the students had rightly received the acclamation they deserved

— Demis Hassabis (@demishassabis) July 21, 2025

Thang Luong, senior researcher at Google DeepMind, stated, “The IMO organizers have their grading guideline. So any evaluation that’s not based on that guideline could not make any claim about gold-medal level performance.”

OpenAI, for its part, says it used three former IMO medalists as independent graders and only released its announcement after the official award ceremony, as advised by IMO. Noam Brown, an OpenAI researcher, noted that the company wasn’t aware Google had been working on an informal AI test with IMO.

IMO officials have not publicly commented, leaving the industry to draw its conclusions.

Google and OpenAI both ranked 27th at the IMO

That’s what I call a heated competition holy moly pic.twitter.com/vHwGKsfdfx

— Chubby♨️ (@kimmonismus) July 22, 2025

The debate may seem petty on the surface — two AI giants arguing over grading and timing — but it reflects a deeper competition for prestige, credibility, and the talent war in AI research.

Many top AI researchers come from competitive math backgrounds, and outperforming at events like IMO sends a strong signal of technical leadership.

Other tech giants are also advancing rapidly in AI. Microsoft recently unveiled Copilot Vision, enhancing visual reasoning across Office apps, while startups like Le Chat are exploring voice-powered AI assistants to push natural interaction even further. Meanwhile, Google is helping students catch up through its AI-powered learning notebooks.

“OpenAI OR Google (or both) just achieved Gold in the International Math Olympiad… but who?” https://t.co/ku7BMec9HZ Grant Harvey (via @theneurondaily) pic.twitter.com/Kz0JoaPRNx

— Stefan Wolpers (@StefanW) July 21, 2025

A Glimpse at the Future of AI Reasoning

Whether or not the grading drama matters, the real takeaway is this: OpenAI and Google’s AI models are closing in on expert-level reasoning. Not just in predictable tasks, but in open-ended problem solving — the kind of thinking that underpins science, engineering, and creativity.

The models didn’t just memorize answers. They reasoned, constructed proofs, and explained their logic in natural language — something AI has traditionally struggled with.

That kind of progress means future AI models might soon be able to tackle not just math problems, but also complex research, planning, and even philosophical debates.

The momentum is building across the board. We’ve already seen new breakthroughs in AI-first browsing tools like Comet, which promise to rethink how users interact with the internet. These innovations point to a future where AI reasoning is integrated across platforms and daily workflows.

And yet, this race is far from over. While OpenAI once held a clear lead in the industry, Google’s recent moves suggest it’s closing the gap fast. With GPT-5 expected later this year, OpenAI is under pressure to maintain its edge — or at least the perception of it.

In the world of AI, perception matters. Talent flows toward the company seen as the most advanced. So even when two AI systems perform equally well, how and when you announce results can shift the balance.

For now, the math speaks for itself. And AI just earned itself a gold.