How do AI models like Celia and dots-note-3.0 compare to human IMO performance?

Question

Accepted Answer

AI Models Achieve Perfect Scores at International Mathematical Olympiad

Artificial intelligence models from Huawei, named Celia, and Xiaohongshu, known as dots-note-3.0, have achieved perfect scores of 100% on the International Mathematical Olympiad (IMO) for the first time. This marks a significant milestone, as previously no large language model had attained a perfect score under the IMO's official judging process. In contrast to this year's AI success, in 2025, AI models from companies like Google and OpenAI reached gold-level scores but did not match the performance of the top human contestants.
AI Models Achieve Unprecedented Perfect Scores

Huawei's AI system, Celia, demonstrated "comprehensive problem-solving capabilities" to achieve a perfect score. Similarly, Xiaohongshu's model, dots-note-3.0, which entered the IMO for the first time this year, also secured full marks. These AI models were presented with the IMO problems after human participants and were required to submit their solutions within a set time limit, with no human intervention permitted during the testing phase.
Human Performance at the 2024 IMO

At the 2024 IMO, held in Shanghai, a total of 666 contestants participated. Among these human competitors, only seven individuals managed to achieve full marks, highlighting the immense difficulty of attaining a perfect score in this prestigious mathematics contest. Participants at the IMO are generally under the age of 20, representing the brightest young mathematical minds globally.
Previous AI Performance and Benchmarking

While this year represents a breakthrough for AI, previous attempts have shown progress. At the 2024 IMO, Google's AI achieved a silver-medal score, successfully solving four out of the six problems presented over the course of the competition. Additionally, four other advanced AI models, including those from OpenAI, Anthropic, Axiom Math, and Moonshot AI's "Kimi K3," each scored a perfect 42 out of 42 points on this year's IMO questions when independently tested.

Reading history

Hide →

This history is cleared when the page is closed.

How do AI models like Celia and dots-note-3.0 compare to human IMO performance?

Loading

Was this article helpful?

Helpful

Not helpful

Report an issue

2

0

0

Share with friends