China’s Newest AI Model Triggers ‘Code Red’ For American AI

China unveiled a new artificial intelligence (AI) model Thursday that seemingly wipes America’s most advanced models at a fraction of the cost.

Moonshot AI, a Beijing-based AI company, released Kimi K3, which now threatens the United States’ lead in the global race on AI. The release of this advanced AI model follows as Anthropic has accused Chinese AI labs of using American AI models to train Chinese AI.

Kimi K3 now ranks first ahead of other top tier of advanced AI models for front-end coding, besting Anthropic’s Fable 5 and OpenAI’s GPT-5.6 Sol in front-end coding tests set by AI testing platform Arena.

Huawei executives designed the Ascend 950PR chip and other advanced chips to work with the MXFP8 data format. This means that the Kimi K3 model could use Chinese-made AI chips that would not be subject to American export controls. the Trump administration oversees the export of Nvidia’s H200 AI chip.

Kimi K3 finished ahead of Anthropic’s Opus 4.8, the company’s top model until the company released Fable 5 in June.

“The entire game has changed. I expect this will trigger some code red for some,” AI expert Kim Isenberg wrote on X.

Moonshot will release its AI product as an open-weight model on July 27, which would allow companies and governments to tailor the model to their preferred specifications.

In April, the National Institute of Standards and Technology (NIST) found that the Chinese AI firm DeepSeek was roughly eight months behind Americans AI models.

However, the release of Kimi K3 appears to have upended America’s edge against China.

Kimi K3’s edge against American AI models may not just be its raw performance. AI clients may prefer a model that nearly has the performance of a top-tier model, especially if it costs 40% percent less and can be customized or run on-site.

Moonshot AI may serve as an existential crisis for Anthropic and OpenAI, both of which have nearly a $1 trillion evaluation based on their technical edge against other AI competitors.

When reached for comment, an Anthropic spokesperson pointed the Daily Caller News Foundation to an Anthropic press release detailing how Anthropic accused three Chinese AI labs, DeepSeek, Moonshot, and MiniMax of “illicitly” extracting the company’s AI models to improve their own modeling using a process called “distillation.” This process involves training a less capable model based on the outputs of a more sophisticated model. The AI company said it is using advanced detection systems to detect and stop distillation attacks.

In April, Chinese Foreign Ministry spokesperson Guo Jiakun denied accusations that Chinese companies use distillation to advance Chinese AI models.

“We urge the US side to respect facts, abandon prejudice, stop the technological suppression of China, and do more things that are conducive to scientific and technological exchange and cooperation between the two countries,” the Chinese government spokesperson said.

The Chinese Embassy in Washington, D.C., in July said that the Trump administrations accusation that Chinese companies are using distilling to better compete against American AI models are “pure slander.”

“China has always been committed to promoting scientific and technological progress through co-operation and healthy competition,” Liu Pengyu, the Chinese embassy spokesperson, told the Financial Times.

“China attaches great importance to the protection of intellectual property rights,” he continued.

The Anthropic spokesperson also referred the DCNF to a letter the company sent to the Senate Banking Committee in June, denouncing Chinese tech company Alibaba for allegedly “brazenly” and “illicitly” trying to extract its AI capabilities.

“Alibaba executed the largest known distillation attack on Anthropic to date,” the letter, which was viewed by the DCNF, stated.

Anthropic, in its letter to the Senate Banking Committee, urged Congress to advance legislation to stop distillation attacks and strengthen export controls. It also advocated for leveraging U.S. national security tools to punish Chinese AI labs that conduct these attacks.

An OpenAI spokesperson did not respond to a DCNF request for comment.

Alibaba, DeepSeek, MiniMax and Moonshot did not respond to a DCNF request for comment.