Key facts
- Chinese AI models from startups like Moonshot and Z.ai are gaining popularity in the US due to lower costs and high efficiency.
- US tech executives and companies are adopting these models for tasks ranging from daily activities to cost reduction.
- Accusations of illicit technology extraction have been made by US entities against Chinese AI startups, which China denies.
- The open-source nature of many Chinese AI models is contributing to their global adoption.
- While competitive, Chinese AI models generally still trail top US counterparts in overall capabilities.
Chinese artificial intelligence models are increasingly making inroads into the US market, offering a compelling combination of lower costs and high efficiency that appeals to developers and companies alike. Models from startups such as Moonshot (Kimi K3) and Z.ai (GLM-5.2) are being adopted for a range of daily tasks, with users like Raffi Krikorian, CTO of Mozilla, finding them 'snappier' and more affordable than established US alternatives like Anthropic's Claude Fable.
This trend is driven partly by the growing demand for cost-effective AI solutions, especially with the rise of 'agentic' AI that requires autonomous, multi-step task execution. Goldman Sachs noted in a July report that Chinese AI models are reaching a critical stage for wide adoption. Analysts highlight that the cost per million tokens for Chinese models can be significantly lower than their US counterparts, making them a practical choice for many users who do not require the absolute cutting edge of AI capabilities.
While Chinese AI models are rapidly advancing and gaining popularity, with platforms like OpenRouter showing top-performing Chinese models and Kimi experiencing a surge in downloads, they still generally lag behind the most advanced US AI leaders in overall capabilities. This has led to accusations from some US politicians and AI companies, including Anthropic, that Chinese startups are illicitly 'distilling' their technologies, claims that Beijing rejects as baseless. US Treasury Secretary Scott Bessent has also warned of potential further sanctions.
Adding to the dynamic, many Chinese AI models are open-source, allowing for broader examination and development, which contrasts with the closed-source nature of leading US models from OpenAI and Anthropic. This open-source advantage is expected to further promote their global usage. In response to the evolving landscape, a group of American tech firms, including Microsoft, Meta, and Nvidia, have publicly backed open AI models.