CISO Voice

AI & Security / News

China’s Moonshot unveils Kimi K3 AI model, closing the gap with US rivals

Moonshot calls it the “most capable” open-weight model.

Akshaya Asokan
Akshaya Asokan Cybersecurity Reporter · July 17, 2026 · 3 min read
Moonshot

Chinese company Moonshot unveiled the company’s “most capable” AI model, which it claims is on par with Anthropic and OpenAI’s latest models.

Moonshot AI is an artificial intelligence company based in Beijing. The company on Thursday unveiled its latest model called Kimi K3, which is set to be released on July 27.

The company said the Kimi K3 is the first open-weight model to reach 2.8 trillion parameters, or learned values, meaning that it will be the largest open-weight model to-date. Because it is an open-weight model, users can iterate and customize it according to their needs.

Moonshot added that the model could be used for frontier intelligence across long-horizon coding, knowledge work, and reasoning. “Its overall performance still trails the most powerful proprietary models, Claude Fable 5 and GPT 5.6 Sol,” the company said.

For security leaders, frontier-grade capability in a cheap open-weight model hands more power to attackers, piling more pressure on vulnerability teams already racing machine-speed exploitation. 

Arena AI, a crowdsourced platform that allows users to test and rank different artificial intelligence models, said on Thursday that Kimi K3 is ranked number one on the platform. “When its output was compared head-to-head against other models on the same task, it was picked as the better output 76% of the time on average.”

Citing industry sources, The Financial Times on Thursday reported that K3 exceeds Anthropic’s Opus 4.8 in terms of capabilities.

Independent software engineer and AI expert Simon Willison, who tested the K3 application programming interface made available by Moonshot, said the model’s vision and text-generation capabilities are good. However, he added that the cost and reasoning estimate for a simple task is “expensive.”

Pricing around closed-source models made by Anthropic and OpenAI has been criticised by industry experts including Microsoft CEO Satya Nadella, who has called it a “reverse information paradox,” as these companies receive the double benefit of getting paid for their service while also gaining access to critical client information.

These arguments are likely to strengthen the case for open-weight models such as K3, as they allow private deployment and scaling, while giving the users better agency and control over the models. Moonshot in May raised around $2B at a $20B valuation, indicating growing value for open-weight AI models in China. 

Increasing Competition

K3 is the latest model released by a company that is on par with leading American AI models. Last month, Beijing-based Zhipu AI released GLM-5.2, another open-weight model that rivalled Anthropic’s Claude Opus 4.8 model.

With the capability gap narrowing, Chinese and American AI sectors have increasingly collided over accusations of spying and intellectual property theft. In February Anthropic accused Moonshot of harvesting Claude’s outputs by using thousands of fake accounts to extract its capabilities to help train its own models.

In turn, the Chinese Ministry of Industry and Information Technology’s Network Security Threat and Vulnerability Information Sharing Platform urged Chinese users this month to not use Anthropic’s Claude over surveillance concerns.

Akshaya Asokan
Akshaya Asokan Cybersecurity Reporter

Akshaya Asokan is a U.K.-based cybersecurity and technology journalist with experience covering nation‑state hacking operations, cyber and tech policy, and emerging cybercrime trends. She previously served as the senior European cybersecurity correspondent for Information Security Media Group’s Global News Desk. Before that, she reported on IT developments for IDG Media and covered artificial intelligence and machine learning for Analytics India Magazine. Earlier in her career, she wrote on sexual and religious minority rights as a reporter for The New Indian Express.

Keep reading for free,
one signup unlocks everything

We'll send you the CISO Voice newsletter. Unsubscribe anytime.

More from CISO Voice

Sign up to CISO Voice, the free newsletter trusted by senior security leaders.

Secret Link