Cheap and effective! American companies increasingly prefer Chinese AI, with OpenRouter platform's share once reaching 46%.

```

Chinese AI models are rapidly penetrating the US enterprise market by offering prices significantly lower than their American counterparts. The advantage of cost-effectiveness, combined with the models’ rapidly improving performance, is pushing more and more US developers and companies to switch their workloads to Chinese models.

On the developer platform OpenRouter, the proportion of tokens generated by Chinese AI models used by US companies has stayed above 30% every week since February 8, once peaking at 46%. By comparison, the average over the previous 12 months was only 11%, and in the first half of 2025, it was even lower at 4.5%. This sharp change has occurred as leading US AI labs like OpenAI and Anthropic continue to raise the prices of their flagship models, driving cost-conscious companies to accelerate the switch.

Cost Pressure Drives the Switch

Price is the core driving force behind this migration wave. According to Justin Summerville from OpenRouter’s data and analytics team, the usage cost of Chinese open-source models is "60% to 90% cheaper" than Anthropic’s and OpenAI’s leading models.

AI startup Lindy is a typical case of this trend. The company switched all its traffic from Anthropic's Claude model to DeepSeek in June 2026. "You can see that cost curve just collapse straight to the floor," Lindy CEO Flo Crivello told CNBC. He expects this decision to save the company millions of dollars within a few months.

Kyle Chan, a fellow at the John L. Thornton China Center of the Brookings Institution, noted, "With AI costs skyrocketing, Chinese AI models have become significantly more attractive to US firms. Previously, American companies didn’t care about the origin of AI models; now, they are becoming much more cost-aware."

Performance Rapidly Catching Up

Beyond the price advantage, the performance of Chinese models is also catching market attention.

GLM 5.2, released by Zhipu in June, scored less than one percentage point below Anthropic’s Opus 4.8 on a widely watched agent benchmark test, yet its cost was only about one-fifth of the latter. Some researchers state that GLM 5.2’s performance on certain cybersecurity benchmarks is now on par with top US labs.

Harpreet Arora, head of agent infrastructure at Vercel, said that GLM 5.2 was adopted at the fastest rate ever tracked by Vercel in 2026 — "during the first full week after launch, daily token volume grew about 27 times and the number of user customers grew about 80 times."

According to Vercel data, DeepSeek’s gateway token share also continued to climb between May and June.

Summerville notes that the new wave of open-source models are "performing excellently, having demonstrated practical capabilities for all but the most complex large language model tasks." Chan estimates that China’s top models are now "about six to nine months behind" the American frontier, but are already competitive enough in many mainstream applications.

Broader Enterprise Adoption

Chinese models are spreading from startups to a wider range of enterprises.

On the AI agent platform LaunchLemonade, targeted at regulated industries, Claude and ChatGPT still top the usage charts, but GLM 5.2 has already broken into the platform’s top five. LaunchLemonade founder and CEO Cien Solon said, "Chinese models such as Zhipu and Alibaba’s Qwen are becoming enterprise options, offering an ideal combination of performance and cost on specific workloads. Enterprises with more mature AI strategies are increasingly willing to use them in scenarios where it makes technical or business sense."

Yacine Jernite, Head of Machine Learning at Hugging Face, underlined a deeper structural trend: "We’re seeing companies increasingly want to shift toward lower cost, more controllable and customizable AI stacks, and given the current state of open-source and open-weight models, that often means using Chinese models."

Risk Disclaimer and Legal NoticeThe market involves risks, and investment should be undertaken cautiously. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situations, or needs of any particular user. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular circumstances. Investing based on this article is at your own risk. ```