AI models from OpenAI and Google DeepMind have announced that their systems earned gold-medal scores in the 2025 International Math Olympiad (IMO).
This competition is one of the most prestigious and challenging mathematics contests for high school students.
The news reflects the increasing power of advanced AI and highlights the tight competition between two of the world’s leading AI companies.
While both models impressed, a debate has emerged over how and when the results were shared.
The IMO
The IMO is no ordinary exam; it challenges the brightest teenage mathematicians from around the world.
Students spend two days solving six complex problems that require not just knowledge, but deep reasoning and creativity.
When competing last year, Google used a “formal” system. This method required humans to rewrite math problems into a format the AI could understand.
In contrast, this year both Google and OpenAI submitted “informal” models. These models read and solved problems in natural language. No human translation was needed.
Both companies reported that their models solved five out of six questions. That score would place them among the top-performing students in the competition.
Also read: Humanity’s Last Exam: The One Test AI Couldn’t Beat
Timing and Transparency
A disagreement has surfaced. OpenAI shared its results shortly after the IMO concluded. However, Google took issue with this announcement.
Thang Luong, a senior researcher at Google DeepMind, criticized OpenAI’s approach.
He said Google waited until Monday morning to share its results out of respect for the student participants.
Google also waited for official grading and approval from the IMO president. But OpenAI used three former IMO medalists to evaluate its AI model.
According to OpenAI, these experts followed IMO’s grading standards. After confirming the gold-level performance, OpenAI contacted IMO.
The organization reportedly asked them to wait until after Friday night’s award ceremony. OpenAI complied, but announced the results early Saturday morning.
The Race
This competition is a reflection of a large-scale rivalry in the AI industry. OpenAI and Google are locked in a struggle to prove leadership.
A benchmark like the AI displays that the company’s AI is beginning to master complex, abstract reasoning.
And success factors like talent, funding, and public trust often depend on how each company is perceived.
