Kimi K3 officially launched on Hugging Face last night, and a rather ironic situation quickly unfolded — while the U.S. government and companies like Anthropic are spending large amounts of money lobbying to block Chinese open-source large models, at least 18 U.S. companies immediately deployed it and offered commercial services. This list compiled by netizen Ding exposed the rift between Washington's political narrative and Silicon Valley's practical reality.

Top performance combined with extreme cost-effectiveness, commercial rationality outweighs political statements

The real logic is not complicated: K3 is currently recognized as one of the world's three strongest models, while the models from OpenAI and Anthropic are closed-source, expensive, and come with many restrictions. Open-source deployment allows companies to be free from reliance on closed-source APIs, enabling them to customize and optimize themselves, significantly reducing usage costs.

Naturally, deploying K3 is far from "download and use immediately." Although this 2.8 trillion parameter giant can run with as few as 8 AMD or Nvidia AI GPUs, Moonshot AI officially recommends a 64-GPU cluster solution. External estimates suggest that the minimum cost for smooth API service is around tens of millions of yuan, and achieving a commercial-level smooth experience requires an investment of about 30 million yuan.

In other words, among these 18 U.S. companies that acted quickly, each has carefully calculated the costs — rather than endlessly paying token-based fees to closed-source vendors, it's better to invest in infrastructure costs once and truly take control of the model's usage rights. In the second half of the commercialization of large models, controlling inference costs is becoming a more important strategic decision than signing exclusive partnerships.