Humans outperformed generative AI models developed by Google and OpenAI at a prestigious international mathematics competition, even though the programs achieved gold-level scores for the first time.
Neither AI model attained full marks — in contrast to five individuals at the International Mathematical Olympiad (IMO), a renowned annual event for participants under the age of 20.
On Monday, Google announced that an enhanced version of its Gemini chatbot successfully solved five out of the six mathematics problems presented at the IMO, which took place this month in Queensland, Australia.
“We can confirm that Google DeepMind has reached the highly sought-after milestone, earning 35 out of a possible 42 points — a score worthy of a gold medal,” the US technology giant quoted IMO president Gregor Dolinar as stating.
“Their solutions were remarkable in numerous ways. IMO graders found them to be clear, precise, and mostly easy to follow.”
Approximately 10 percent of human participants earned gold-level medals, with five achieving perfect scores of 42 points.
OpenAI, the creator of ChatGPT, reported that its experimental reasoning model scored a gold-level 35 points on the examination.
The outcome “accomplished a long-standing grand challenge in AI” at “the world’s most esteemed math competition,” OpenAI researcher Alexander Wei shared on social media.
“We assessed our models on the 2025 IMO problems under the same conditions as human participants,” he explained.
Last year, Google secured a silver medal at the IMO held in Bath, UK, solving four out of the six problems.
This process required two to three days of computation — significantly longer than this year, when its Gemini model resolved the problems within the 4.5-hour time limit, as stated by the company.
Related News: http://Rivers Farmers Harness AI to Combat Climate Challenges
The IMO indicated that technology firms had “privately tested closed-source AI models on this year’s problems,” which were also faced by 641 competing students from 112 nations.
“It is incredibly exciting to witness advancements in the mathematical abilities of AI models,” remarked IMO president Dolinar.
Contest organizers were unable to verify the extent of computing power utilized by human involvement, he cautioned.
