Ai Openai and Google Deepmind Gold Medal points reached the 2025 International Math Olympic Ado (IMO), one of the world’s oldest and most challenging high school mathematics competitions, companies independently announced in recent days.
The result emphasizes how fast the AI system is progressing, and yet how steadily the Google and Openai seem to be in the AI competition. AI companies are fiercely competing for the audience’s perception of being in the AI competition: a battle for intangible “sound” that can have a major impact on safeguarding the best abilities of AI. Many AI researchers come from the background of competitive mathematics, so benchmarks like IMO mean more than others.
Last year, Google made a silver medal in IMO Using the “formal” system, meaning that it required people to turn problems into a machine gun. This year, both Openai and Google participated in unofficial systems that were able to enjoy questions and provide evidence -based answers in natural language. Both companies claim that the AI models were correctly answered by five six IMO test issues, and the scoring was higher than most high school students and Google AI model from last year without demanding a human machine translation.
In TechCrunch interviews, Openain and Google IMO effort researchers claimed that these gold medal shows represent breakthroughs in unicideous areas around the AI. While AI -concluding models usually have a good deal with straightforward answers, such as simple mathematics or coding tasks, these systems are struggling in tasks that have unclear solutions such as buying a large chair or helping with complex research.
However, Google raises questions about how Openai performed and reported the gold medal IMO performance. After all, if you are going to participate in AI models in a university mathematics competition, you may just as well claim like teenagers.
Shortly after Openai announced feat Saturday morning, CEO of Google Deepmind and researchers took the social media Slam Openai when announcing her gold format prematurely – Shortly after IMO announced which universities had won the competition on Friday night – and that the model test by IMO was not officially evaluated.
Thang Luong, Google Deepmind researcher and IMO project, told Techcrunch that Google expected to announce its IMO results to honor students in the competition.
TechCrunch event
San Francisco
And
27.-29. October 2025
Luong said that Google has worked with IMO organizers from last year to test and wanted to get the IMO President’s blessing and official classification before Indicating the official results it made on Monday morning.
“IMO organizers have a classification guide,” Luong said. “So any evaluation that is not based on the instructions could not make requirements for the gold medal level (performance).”
Noam Brown, an older Openai researcher at the IMO model, told Techcrunch that IMO reached Openaii’s participation in a formal mathematics competition a few months ago, but the chatgpt manufacturer refused because it worked on natural language systems who thought it was more worth practicing. Brown says Openai did not know that IMO conducted an informal test with Google.
Openai says it has hired a third-party evaluators-three former IMO medalist who understood the classification system to classify the performance of the AI model. When Openai learned about his gold medal points, Brown said the company contacted IMO: Hon, who then ordered the company to wait for announcing IMO Friday night after the award ceremony.
IMO did not respond to Techcrunch’s comment request.
Google is not necessarily wrong here – it goes through a more official, tougher process to achieve its gold medal score – but the conversation may forget a bigger picture: AI models of several leading AI laboratories are quickly healed. The countries around the world sent their brightest students to compete in IMO this year, and only a few percent of them made both Openain and Google AI models.
Although Openai was in a significant lead in the industry in the past, it certainly seems that the competition would be more accurate than any company would like to admit. Openain is expected to publish GPT-5 in the coming months, and the company will surely hope to give the impression that it will continue to lead the AI industry.
