Chinese AI models score 100% at the world’s hardest student math contest, and the result is a clear signal that the global AI race just moved up a gear. On July 22, 2026, two Chinese tech companies, Huawei and Xiaohongshu, announced that their AI systems each solved every problem at the International Mathematical Olympiad (IMO), held this year in Shanghai. No AI had ever done this before under the IMO’s official rules. For countries like Pakistan that are building their own AI strategies, this moment matters a great deal.
What Happened at IMO 2026
The IMO is the world’s most respected math competition for pre-university students. This year, 666 contestants from different countries took part in Shanghai. Getting a perfect score is rare even for the best human mathematicians. A perfect score at the contest used to be the preserve of a handful of bright young minds each year, but for the first time, artificial intelligence has also earned a perfect score.
Huawei said its AI system “Celia” had shown “comprehensive problem-solving capabilities” across several mathematical fields. Xiaohongshu, known as RedNote in English, said its model “dots-note-3.0” had entered the IMO for the first time this year. Both models solved all six problems correctly.
The process was strict. Tech firms were only provided with the IMO problems after human contestants had taken the exam, with their submissions required within a specified time limit. “During testing, any form of human intervention was strictly prohibited,” and the AI solutions were submitted to IMO organisers for grading, Xiaohongshu said. In short, no one helped the machines. They did it alone, on a clock.
How Fast Has AI Come on the IMO
The speed of AI’s progress on this benchmark is striking. At the 2024 IMO, Google achieved a silver-medal score, solving four of the six problems over two to three days. In 2025, models made by companies such as Google and OpenAI reached gold-medal scores for the first time, but could not match the abilities of five human contestants who earned 100%. Now, just one year later, Chinese AI models score 100% and match the best humans outright.
Huawei and Xiaohongshu were the first AI labs to announce their scores, but others from around the world may follow suit in the coming days. That means we may soon see even more labs, from the US, Europe, and elsewhere, claiming the same achievement. The ceiling keeps rising.
Why This Signals China’s Growing AI Strength
Huawei is under heavy US sanctions that cut it off from advanced chips and American software. Yet its AI system still solved every IMO problem perfectly. Xiaohongshu is best known as a social platform (like Instagram for China), yet its research arm built a model that passed the world’s toughest student math test on its first attempt. This tells us that China’s AI talent and research culture runs deep, well beyond the names you might expect.
This is not an isolated moment. It follows Chinese AI firm DeepSeek releasing open-source math models that also scored at gold level. DeepSeek’s accomplishment was expected to lower the barrier for developers around the world to access a powerful maths AI model, a few months after leading US AI giants Google DeepMind and OpenAI achieved the same feat at IMO with their proprietary models. China is no longer chasing the US in AI, it is competing head to head, and in some areas, moving ahead.
What This Means for Pakistan’s AI Ambitions
Pakistan launched its National AI Policy 2025 just last year. The policy sets bold targets. It aims to generate AI awareness programs, incorporate AI concepts from early phases of STEM education, and upskill both youth and working professionals through mass training initiatives. Time-bound human-capital targets include 90% awareness by 2026, one million learners and 10,000 trainers by 2027, plus public-sector upskilling and scholarship uplift.
Those are good goals. But the IMO milestone puts them in sharp relief. Pakistan’s policy is largely focused on broad awareness and training volume. What the IMO result shows is that the countries pulling ahead are investing in deep mathematical reasoning, the kind of research that lets an AI solve six brutally hard problems in a row without any human help.
There are also real structural gaps. Pakistan is experiencing a historic brain drain of technology professionals to the GCC and Europe. The policy emphasises talent creation while ignoring retention and repatriation. Training one million AI professionals is a headline metric, but mass certification means little if the talent leaves for higher-paying jobs abroad.
Pakistan does have a starting point. The government has already established nine Centers of Excellence in AI and related fields in major universities to build the talent pipeline. Pakistan’s youth regularly compete in international math olympiads and produce strong results. The raw talent is there. The question is whether the system around that talent, funding, compute, research labs, and a reason to stay in Pakistan, can grow fast enough to keep pace with what China and the US are doing. The government’s $20M Pakistan Venture Fund for Robotics and AI is a step in the right direction, but the scale of investment needed is far larger.
What Comes Next in the AI Math Race
The IMO perfect score is a research milestone, not a product. But research milestones matter because they show which countries are building AI that can truly reason, not just predict the next word. AI that can do olympiad-level math can also help design drugs, write better code, solve engineering problems, and run financial models. This is the practical value hiding behind the headlines.
For Pakistan, the IMO result should act as a reality check. The global bar is rising fast. Having a national AI policy written on paper is only the first step. The real test is whether Pakistan can produce, fund, and keep the researchers who will build the next generation of reasoning AI, and whether the country’s students who love math have a reason to build that future at home.
Frequently Asked Questions
Which AI models scored 100% at IMO 2026?
Two Chinese AI models, Huawei’s “Celia” and Xiaohongshu’s “dots-note-3.0”, each scored 100% on the six problems given to human contestants at the 2026 International Mathematical Olympiad in Shanghai. They were the first AI labs to announce perfect scores at this year’s contest.
Has any AI ever scored 100% at the IMO before?
No. This is a first. In 2024, Google’s AI reached a silver-medal level. In 2025, Google and OpenAI reached gold-medal level but still fell short of a perfect score. The 2026 result, achieved by Chinese AI models score 100% under strict, no-intervention conditions, is a new high point for AI mathematical reasoning.
What is the IMO and why does it matter for AI?
The International Mathematical Olympiad is an annual competition for pre-university students. It tests very deep problem-solving, not just fast calculation. When an AI can score 100% at the IMO, it shows that the system can reason step by step through complex, multi-stage problems, a skill that has many real-world applications in science, engineering, and software.
How should Pakistan respond to China’s AI progress?
Pakistan’s National AI Policy 2025 sets training targets and builds AI centers at universities, which are positive steps. But experts point out the policy needs stronger focus on keeping trained talent inside the country, building research capacity for deep mathematical and scientific AI, and securing the compute infrastructure needed to train advanced models locally.
