The AI has just achieved a great success in international mathematics.
For the first time, Google Deep Mind and Open AI’s AI models Gained the status of a gold medal At the International Mathematics Olympiad (IMO) of 2025, a difficult math competition for high school students that has been held annually From 1959. Comparison Is included Two 4.5 hours of exams to solve six total problems without the help of the Internet or external tools.
Related: Google’s AI Initiative CEO is worried about 2 things, nor is AI replaced by jobs
New York Times There are reports that Openi and Google’s AI models answered questions without any human intervention using natural language. Both models managed to solve five of the five issues offered against 2025, which for the first time, the first time AI models achieved the level of success.
Both models tied in score, each scored 35 points out of possible 42 points on the IMO, exactly fine Cut off point For gold medals. Open I Announced The results of the week when Google waited until Monday.
Google Deep Mind worked with the IMO to classify and verify the performance of its AI system by the committee this year, while the Open did not formally enter the competition. Instead, the Openi asked three former IMO medals to freely classify the answers to every question of his AI model, then finalize the score. “Agreed agreed“
According to Google DeclarationOnly 8 % of the high school students competing in the IMO get a gold medal. Google’s Gold Medal’s performance this year was one step above the results last year, when its AI Found a silver medalSolving four of the six issues offered in the competition.
Related: How chess love attracted Google’s Deep Mind’s CEO to a career in AI – and a Nobel Prize
Dolinar said that IMO President, Dr. Gregor Dulner, has called the Google Deep Mind solution this year “surprised in many cases”, while the IMO’s Graders (the solution) found “clear, exactly, and most of them easy to follow.”
Sam Altman, CEO of Openi In a post on x On Saturday, the opening “has not intended to release the AI model with the capabilities of the IMO Gold for several months, the” gold medal “was an important sign of the extent to which the AI has come in the past decade.” Openi used a general purpose reasoning system to deal with the competition, not a special math system, as the company works towards common intelligence.
We achieved gold medal level performance with a general purpose reasoning system compared to 2025 IMO! To emphasize, this is an LLM that is doing mathematics and not a specific formal math system. It is part of our major push towards general intelligence.
When we first started an open, … https://t.co/x46rspi4l6
– Sam Altman (@SAMA) July 19, 2025
Meanwhile, Google Deep Mind’s CEO Demis Huzzi wrote In a post on x On Monday, Google also used an innovative version of its general -purpose Gemini reasoning model, which will be available for “a set of reliable testers” before paying Google A. Ultra users. Month 250 every month For advanced capabilities and 30 TB storage.
We achieved this year’s impressive results using a modern version of the Gemini Deep Think (complex issues). Our model worked in natural language from end to end, which developed proof of strict mathematics directly in detail of the official problem.
– Dames calculation (@Demishabis) July 21, 2025
This year, 630 high school students participated in the IMO in Queensland, Australia, in which 67 students received gold medals, Reuters.
I join top CEOs, founders, and operators Level -up conference Unlock strategies to scal your business, increase revenue and build sustainable success.