China’s Open AI Brands Are Challenging Silicon Valley’s Playbook

AI companies does not meet a DeepSeek 2.0 minutesbut I feel very close. The director Chinese AI laboratories has been on a roll recently, releasing a series of almost cutting-edge open-source models. Z.ai released GLM 5.2 in June, Moonshot AI released Kimi K3 last week, and Alibaba released Qwen 3.8 on Monday.

Silicon Valley and Washington started talking about models right away, especially K3, which is known to be the best. David Sacks, a venture capitalist and AI adviser to President Donald Trump, called the system “Moonshot”.about.” Earlier this week, Commerce Secretary Scott Bessent encouraged the US can force sanctions on Chinese AI companies.

On Wednesday, Michael Kratsios, director of the White House Office of Science and Technology Policy, he says that the Trump administration has “information that Moonshot AI tampered with Anthropic’s Fable to create its own version of K3,” which he said was “stealing US technology and destroying American research” and was “unacceptable.” (Moonshot AI did not immediately respond to a request for comment.)

The new Chinese models have a number of things in common: third-party benchmarks show that they work just as well as the best Western models; they are designed for coding (the hottest thing in AI this year); and they are or will soon be released with open weights, making them accessible and visible.

But perhaps the biggest similarity between the current time and January 2025 – when the world was shocked by the R1 version of DeepSeek – is that it also confirms how American and Chinese AI labs are taking revolutionary steps when it comes to opening or closing.

When it first exploded, DeepSeek challenged the idea that only closed brands built with billions of dollars in computer hardware and education could succeed. But since then, Western AI labs have continued to develop AI in the same way, and now America’s frontier species are feeling more wired than they were a year ago.

Anthropic said for months that its latest version of Mythos was the best at hacking Only authorized service providers he can use it. When more was released, the White House responded by releasing more information external controlswhich forced Anthropic to take Mythos and its incompetent sister, Fable 5, offline for a while. OpenAI similarly delayed the release of GPT 5.6 later received a request from the White House.

In China, meanwhile, things look very different. Chinese startups and tech giants are outnumbered: Anyone with a decent computer environment can download a lightweight version, run it locally, add customizations, and enjoy more freedom than OpenAI and Anthropic would allow. In many ways, the open versus closed debate is more involved in the US-China conflict than ever before.

There are many reasons why Chinese labs have chosen a business model built on top of open source models. Being a new, small fish in the AI ​​field, making their models free and open can help Chinese companies attract more users, partners, and media. It also puts them on a different footing than OpenAI, Anthropic, Google, SpaceX, and other deep-pocketed giants.

Earlier this year, rumors spread that Alibaba might be considering joining a closed source after the competition reorganized its AI development teams. But the tech giant announced Monday that it will also release the latest version of Qwen — its line of open-source models. loved by the international art community– with open weights, showing them to customers and the public is still not working.