A perfect score at the world's premier youth mathematics competition has historically been achieved by only a tiny selection of human prodigies each year, but artificial intelligence has now reached the same milestone.
Chinese technology companies Huawei and Xiaohongshu announced that their respective artificial intelligence systems scored 100 per cent on the exact problems presented to human participants at this month's International Mathematical Olympiad.
The announcements represent the first confirmed instances of artificial intelligence tools achieving complete accuracy under the formal grading guidelines of the competition.
Speaking in a company statement, representatives from Xiaohongshu confirmed that their system had successfully solved all six complex problems without human help.
The milestone follows significant progress recently, as systems developed by Google and OpenAI achieved gold-medal standards in 2025 without reaching maximum points.
This year, 666 human participants under the age of 20 gathered in Shanghai for the annual competition.
Only seven human contestants managed to score full marks across the entire event, according to official scoreboard records.
Huawei confirmed that its system, known as Celia, demonstrated advanced reasoning capabilities across multiple advanced mathematical disciplines during testing.
Xiaohongshu stated that its model, named dots-note-3.0, competed in the event for the first time under strict timing constraints after the human examination was completed.
Organisers verified that tech firms received the examination papers only after human students had finished, and that all AI-generated solutions were graded by official contest judges without human intervention.
Industry analysts noted that the speed of progress has accelerated rapidly, moving from silver-medal performance levels in 2024 to multiple perfect submissions across leading global tech labs this year.
Menlo Ventures partner Deedy Das noted on social media that several other cutting-edge systems from American and Chinese development teams also scored maximum points when tested independently on this year's examination paper.
The results mark a notable shift in computing capabilities, proving that advanced language models can now master logical reasoning tasks previously considered exclusive to top human minds.
