Artificial intelligence has reached another major milestone after achieving a perfect score at the world’s most prestigious mathematics competition for the first time, matching the performance of the competition’s brightest human contestants.
Chinese technology companies Huawei and Xiaohongshu (RedNote) announced that their AI models each scored 100 percent in this year’s International Mathematical Olympiad (IMO)—a competition widely regarded as one of the toughest mathematics contests in the world.
The companies said their AI systems solved every question correctly under the IMO’s official judging process, making it the first time a large language model has achieved a perfect score.
“We are delighted with this result, because achieving a perfect score at the IMO is extremely challenging,” Xiaohongshu said in a statement.
The company added:
“Previously, no large language model had ever achieved a perfect score under the IMO’s official judging process.”
The 2026 competition featured 666 contestants from countries around the world, but only seven students managed to earn full marks, according to the official IMO scoreboard.
Huawei said its AI model, Celia, demonstrated strong problem-solving abilities across multiple areas of mathematics, while Xiaohongshu revealed that its dots-note-3.0 model participated in the challenge for the first time.
The companies explained that the AI systems only received the exam questions after human contestants had completed the competition, and were required to submit their answers within the official time limit.
“During testing, any form of human intervention was strictly prohibited,” Xiaohongshu said.
The rapid progress of AI has surprised many experts. At the 2024 IMO, Google’s AI achieved only a silver-medal performance by solving four out of six problems. By 2025, AI models from companies including Google and OpenAI reached gold-medal level, but still fell short of achieving a perfect score.
Adding to the excitement, Deedy Das, a partner at Menlo Ventures, said he independently tested this year’s IMO questions on four advanced AI models developed by OpenAI, Anthropic, Axiom Math, and Moonshot AI.
According to him, all four models also achieved the maximum 42 out of 42 points.
“The frontier of AI has officially moved well past IMO math,” Das wrote on LinkedIn.
The achievement highlights how quickly artificial intelligence is advancing and raises fresh questions about its growing ability to solve highly complex problems once considered the exclusive domain of exceptional human talent.
Source: AFP.
