China Daily: Kimi K2 as a DeepSeek moment

mirrored 2026-10-22 · origin

title: "China Daily: Kimi K2 as a DeepSeek moment" source: https://global.chinadaily.com.cn/a/202507/21/WS687dcf34a310ad07b5d9103a.html author: unknown date: unknown models: [kimi-k2] tags: [press] mirrored: 2026-10-22 note: mirrored against link rot by the Pantheon; all rights with the original author

[Global Edition

]()

Global Edition

Home | Technology

Chinese AI model Kimi K2 marks 'another DeepSeek moment': Nature

Xinhua | Updated: 2025-07-21 13:25

[]()

[](http://www.linkedin.com/shareArticle?mini=true&title=Chinese AI model Kimi K2 marks 'another DeepSeek moment': Nature&url=https://global.chinadaily.com.cn/a/202507/21/WS687dcf34a310ad07b5d9103a.html)

[](https://twitter.com/share?text=Chinese AI model Kimi K2 marks 'another DeepSeek moment': Nature&url=https://global.chinadaily.com.cn/a/202507/21/WS687dcf34a310ad07b5d9103a.html)

[](https://www.facebook.com/sharer.php?title=Chinese AI model Kimi K2 marks 'another DeepSeek moment': Nature&u=https://global.chinadaily.com.cn/a/202507/21/WS687dcf34a310ad07b5d9103a.html)

LONDON -- The prestigious scientific journal Nature has hailed China's latest artificial intelligence (AI) model, Kimi K2, as "another DeepSeek moment" for the global AI community.

An article published this week by Nature said that following the significant impact of DeepSeek's R1 launch in January, researchers around the world are now increasingly enthusiastic about the emergence of China's second high-performance AI model.

On July 11, Beijing-based Moonshot AI unveiled Kimi K2, with the company claiming its performance equals or surpasses that of Western rivals, as well as some DeepSeek models, across various benchmarks.

The article highlights that Kimi K2 particularly excels at coding, achieving high scores in tests such as LiveCodeBench, a method for evaluating AIs that challenges models on code-related tasks.

The model also ranks first on the Creative Writing v3 benchmark, which assesses narrative authenticity and originality, and leads the EQ-Bench 3 leaderboard, designed to measure emotional intelligence through complex role-playing scenarios.

According to official data, Kimi K2 has a total parameter count at the trillion scale (1T). However, it employs a "mixture-of-experts" (MoE) architecture, activating only 32 billion parameters at a time.

This design "allows it to use only the relevant parts of the model for each task," Nature explains, thereby reducing the amount of computing power required.

Similar to DeepSeek models, Kimi K2 is open-weight, meaning it can be freely downloaded and modified by researchers. Additionally, it is accessible via an application programming interface (API) at a fraction of the cost of proprietary models such as Claude 4 from Anthropic.

Founded in March 2023, Moonshot AI was relatively unknown in the West. However, by November 2024, its earlier Kimi chatbot had become the third most widely used in China, according to Hong Kong-based market research firm Counterpoint.

The article suggests that the launch of a second high-performance model within six months indicates that China's recent AI breakthroughs are part of a sustained trend rather than isolated successes.

Mario Krenn, head of the Artificial Scientist Lab at the Max Planck Institute for the Science of Light in Erlangen, Germany, told Nature: "I wouldn't be surprised if more will come (from China) in the next months."

Photo

Xi calls for new achievements on new Long March

On the Road to Victory

China holds gathering to mark 90th anniversary of Long March victory

City diplomacy seen as bridge amid global challenges

Shanghai studio brings Chinese icons to the world

Xi Jinping's new Long March

[