Artificial intelligence has reached another milestone in its rapid evolution. For the first time, AI systems have reportedly achieved a perfect score at the International Mathematical Olympiad (IMO), one of the world’s most prestigious and challenging mathematics competitions for high school students.

Chinese technology companies Huawei and Xiaohongshu announced that their large language models (LLMs) solved every problem in this year’s competition, earning a perfect 42 out of 42 points. If independently verified, the achievement would represent a significant advance in AI’s mathematical reasoning capabilities and highlight how quickly frontier AI models are closing the gap with top human problem-solvers.

The breakthrough follows years of steady progress, with AI advancing from silver-medal performance in 2024 to gold-level scores in 2025 before now reportedly achieving perfection.

TL;DR

• Huawei’s AI model “Celia” and Xiaohongshu’s “dots-note-3.0” reportedly achieved a perfect score at the 2026 International Mathematical Olympiad.
• Both models solved all six IMO problems, earning 42 out of 42 points.
• The AI systems received the questions only after the human competition had concluded.
• The achievement marks the first reported instance of a large language model scoring 100% on the IMO.
• The result highlights rapid advances in AI reasoning, with potential implications for scientific research, engineering and education.
• Independent verification of the companies’ claims will be important in assessing the milestone.

What Is the International Mathematical Olympiad?

The International Mathematical Olympiad is widely regarded as the world’s premier mathematics competition for high school students.

Held annually since 1959, the IMO brings together the brightest young mathematicians from more than 100 countries to solve exceptionally difficult proof-based problems.

Unlike standard math exams, the competition emphasizes creativity and logical reasoning rather than memorization or computation.

Contestants have two days to solve six problems covering areas such as:

• Algebra
• Geometry
• Number theory
• Combinatorics

Each problem is worth seven points, giving competitors a maximum possible score of 42.

What Did the AI Models Achieve?

According to Huawei and Xiaohongshu, their AI systems solved every problem presented at the 2026 IMO.

Huawei’s model, Celia, and Xiaohongshu’s dots-note-3.0 reportedly earned full marks by successfully producing correct mathematical solutions for all six problems.

Importantly, the companies said the models were given access to the questions only after the official competition had ended, preventing any possibility of prior exposure during the contest.

If confirmed, this would be the first time a large language model has achieved a perfect score on one of mathematics’ toughest competitions.

Why Is This Achievement Significant?

Mathematical Olympiad problems are fundamentally different from the types of calculations AI systems routinely perform.

They require:

• Multi-step logical reasoning
• Creative problem-solving
• Abstract mathematical thinking
• Rigorous proof construction

Unlike straightforward arithmetic or symbolic computation, IMO problems often demand original insights that even professional mathematicians find challenging.

For years, mathematical reasoning has been considered one of AI’s greatest weaknesses.

A perfect score suggests that leading AI models are becoming increasingly capable of handling complex reasoning tasks that extend beyond pattern recognition.

How Has AI’s IMO Performance Improved?

Progress has accelerated remarkably over the past few years.

2024: Silver-medal performance

Google’s AI system reportedly achieved a silver-medal-level score after solving four of the six IMO problems.

However, completing those solutions reportedly required two to three days.

2025: Gold-level breakthrough

Models developed by Google and OpenAI reached gold-medal performance for the first time.

While impressive, they still fell short of the handful of human contestants who achieved perfect scores.

2026: Perfect score

This year’s reported results represent another leap forward.

Huawei and Xiaohongshu say their models solved all six problems, matching the maximum possible score of 42 out of 42.

Separately, venture capitalist Deedy Das of Menlo Ventures reported testing four frontier AI models using this year’s IMO questions, with all four reportedly achieving perfect scores.

He summarized the milestone by writing on LinkedIn:

“The frontier of AI has officially moved well past IMO math.”

Recent Posts

Need a new Car? Rent To Own Cars No Credit Check

need a new car? rent to own cars no credit check ...

July 27, 2026

9:28 am

2025 Jeep Wrangler Price One Might Not Want to Miss!

2025 jeep wrangler price one might not want to miss!...

July 27, 2026

9:25 am

Celebrate the Holidays in a New Hyundai Palisade

celebrate the holidays in a new hyundai palisade...

July 27, 2026

9:19 am

Explore The 2025 Jeep Compas: Adventure Awaits!

explore the 2025 jeep compas: adventure awaits!...

July 27, 2026

9:39 am

How Do Large Language Models Solve Math Problems?

Modern large language models are trained on vast amounts of text, code and mathematical content.

When solving advanced mathematical questions, they typically combine:

• Logical reasoning
• Pattern recognition
• Step-by-step inference
• Symbolic manipulation

Recent advances in reasoning-focused AI models have significantly improved their ability to tackle complex problems by generating intermediate reasoning steps before arriving at a final answer.

This approach has dramatically narrowed the gap between AI and expert human mathematicians on benchmark tests.

Does This Mean AI Is Better Than Human Mathematicians?

Not necessarily.

While solving IMO problems is an impressive demonstration of reasoning ability, mathematical competitions represent only one dimension of mathematical expertise.

Professional mathematicians also:

• Develop entirely new theories.
• Formulate original research questions.
• Collaborate across disciplines.
• Produce proofs that advance mathematics itself.

Current AI systems excel at solving existing problems but are still being evaluated on their ability to consistently generate genuinely novel mathematical discoveries.

Moreover, independent verification remains important whenever companies announce benchmark achievements.

What Could This Mean for Science and Engineering?

Improved mathematical reasoning could significantly expand AI’s usefulness across scientific disciplines.

Potential applications include:

• Scientific research
• Engineering design
• Physics simulations
• Drug discovery
• Cryptography
• Climate modeling
• Robotics
• Advanced software development

Many scientific breakthroughs depend on solving highly complex mathematical problems.

As AI becomes better at formal reasoning, it could increasingly serve as a research assistant capable of accelerating discovery rather than simply automating routine tasks.

Why Independent Verification Matters

While the reported results are remarkable, benchmark claims are generally strongest when independently validated by third-party researchers or competition organizers.

Key questions include:

• Were the solutions evaluated using official IMO grading standards?
• Were the models allowed any external computational tools?
• How much processing time did they require?
• Were the solutions generated in a single attempt or through multiple iterations?

Independent assessment helps ensure comparisons remain fair across different AI systems and years.

The Bigger Picture for AI

The reported IMO achievement reflects a broader trend in artificial intelligence.

Over the past few years, frontier AI systems have demonstrated rapid improvements in:

• Coding
• Scientific reasoning
• Mathematical problem-solving
• Language understanding
• Image generation
• Research assistance

What once required specialised software increasingly appears achievable using general-purpose AI models capable of reasoning across multiple domains.

Whether these systems can consistently translate benchmark success into real-world scientific breakthroughs remains one of the most closely watched questions in AI research.