appConfigEntityId

bodyType

customContents

entityId

entityUuid

isVideo

articlePageQuery

article

articleSchemaContent

content

hooksArticlePageAdvertisementConfigArticle

hooksArticlePageFloatingButtonConfigArticle

appConfig

filter

json

liveArticle

Richard Chen

For the first time, an artificial intelligence model has achieved a perfect score at the International Mathematical Olympiad (IMO) – and it was developed by Chinese social giant RedNote.
The model, dots-note-3.0, solved all six puzzles and secured a perfect score of 42 at this year’s IMO, which concluded in Shanghai on Monday.
The note variant, still in beta phase, is currently the lightest model in the dots3 family, which includes two larger versions – jazz and aria – tailored for different use cases and compute costs.
Shanghai-based RedNote, known for its Instagram-style social community called Xiaohongshu in Chinese, said it would open-source dots-note-3.0 “in the near future”, without providing further details, according to a post on the model’s website.
Held annually since 1959, the IMO is widely regarded as the world’s most prestigious mathematics competition, including for AI systems. Unlike answer-only exams where only the final results need to be judged, IMO problem-solving requires proof backed up by rigorous arguments, without logical errors or missing conditions.
Earlier attempts by AI have scored fairly well. At last year’s IMO, Google DeepMind and OpenAI achieved gold-medal-level performances, both earning 35 out of 42 points.

At IMO 2024, Google DeepMind’s AlphaProof and AlphaGeometry 2 scored 28, a silver-medal equivalent, but required human experts to manually translate problem statements into logic language for the AI to process them.
This year, RedNote processed the original maths competition documents directly, without the need for humans to translate them into specialised coding languages. The system operated end-to-end under official contest rules that strictly barred any human intervention, such as providing hints, correcting reasoning steps or completing logic.
Dots-note-3.0 relied on an “agentic” framework that combined natural-language reasoning with Python programming code execution to draft, test and refine its proofs.
According to RedNote, the system’s strong performance was rooted in its ability to scrutinise its own reasoning, question assumptions and continuously fix flawed logic through a multi-step self-review loop before submitting final solutions.
Separately, the Chinese human team consisting of six members secured first place at the IMO, with three achieving a full score.


Philippine tribe vows to stand ground against US-led Pax Silica: ‘listen to us’

As the US weighs restrictions on cloud computing, how will China’s AI sector adapt?

Chinese AI chipmaker Hygon plots expansion from data centres to robotics

For China’s hi-tech IPOs, it’s Hong Kong vs the mainland – but how much does it matter?

Artificial intelligence (AI)

World’s first AI model to earn perfect score at maths Olympiad comes from China’s RedNote

The RedNote app is seen on a smartphone in this illustration taken January 15, 2025. Photo: Reuters

An ad for Chinese social app Xiaohongshu, also known as RedNote, seen at a shopping centre in Beijing on January 15, 2025. Photo: AFP

Google and OpenAI’s AI models win milestone gold at International Mathematical Olympiad

Google’s research facility in Mountain View, California. Photo: Reuters

China’s 2025 IMO championship student team. Photo: Handout

DeepSeek releases first open AI model with gold-level scores at maths olympiad

DeepSeek on Thursday made its Math-V2 open-source AI model widely available. Photo: Shutterstock

DeepSeek’s Math-V2 model showed strong theorem-proving capabilities to achieve gold-level scores at the 2025 International Mathematical Olympiad. Photo: Shutterstock

Tech

News

China Future Tech

At last year’s IMO, Google DeepMind and OpenAI achieved gold-medal-level performances, both earning 35 out of 42 points


At last year’s IMO, Google DeepMind and OpenAI achieved gold-medal-level performances, both earning 35 out of 42 points.


World’s first AI model to earn perfect score at maths Olympiad comes from China’s RedNote

.css-1c6uqr6{color:inherit;font-weight:inherit;font-size:inherit;font-family:inherit;line-height:inherit;overflow-wrap:break-word;}At last year’s IMO, Google DeepMind and OpenAI achieved gold-medal-level performances, both earning 35 out of 42 points

At last year’s IMO, Google DeepMind and OpenAI achieved gold-medal-level performances, both earning 35 out of 42 points