Chinese AI Models Gain Ground in the US Market

A quiet but significant shift is underway in the artificial intelligence landscape: Chinese AI models are rapidly gaining traction in the United States, offering cheaper and more open alternatives that challenge the dominance of American tech giants. From San Francisco coders to Fortune 500 companies, a growing number of US users are turning to models from Chinese startups like Moonshot, DeepSeek, and Z.ai — drawn by performance that rivals frontier US systems at a fraction of the cost.

According to the Associated Press, the trend marks a new phase in the US-China AI race, one where cost efficiency and open-source accessibility are becoming as important as raw capability.

The Rise of Affordable AI

The breakthrough moment came in early 2025, when Chinese startup DeepSeek stunned the US tech industry with a model that matched American frontier performance at a dramatically lower cost. Since then, the pace has only accelerated. In June 2026, Z.ai (also known as Zhipu) released GLM-5.2, followed in July by Moonshot AI’s Kimi K3 — a model with 2.8 trillion parameters, making it the largest open-weight AI model ever built.

Alibaba previewed its Qwen3.8 Max model in July as well, while DeepSeek rolled out previews of its V4 model in April. Each new release narrows the gap with US frontier models from OpenAI and Anthropic.

The numbers tell the story. Sensor Tower data cited in the AP report shows Kimi had more than 930,000 downloads worldwide in the week after K3’s release — a 200% increase from the prior week. In the US alone, downloads surged 387% to roughly 86,000. Over the past month, the five most popular models on OpenRouter, a platform tracking AI model usage, were all Chinese.

Why US Users Are Switching

For many developers and businesses, the calculus is simple: Chinese models are “good enough” at a fraction of the price. Curt Meinhold, a North Carolina-based tech executive who founded digital legacy platform LilyList, told AP he prefers DeepSeek for tasks like finding business leads and generating sales.

“At the end of the day, most of us, the vast majority of us, 90 plus percent, don’t need Mythos or Fable,” Meinhold said. “If I can pay a handful of cents per million output tokens versus 30 bucks or 40 bucks or 50 bucks, then it’s good enough.”

Raffi Krikorian, CTO of Mozilla, echoed the sentiment. He switched to Kimi K3 for day-to-day activities within days of its launch, calling it “snappier” than Anthropic’s Claude Fable. He noted that he had already been using Z.ai’s GLM-5.2 for routine tasks like calendar management.

“The open frontier is becoming increasingly Chinese-built,” Krikorian observed.

The Distillation Controversy

The rapid rise has not gone unnoticed in Washington. On July 22, White House science chief Michael Kratsios accused Moonshot of conducting “covert” distillation of Anthropic’s Fable model to build K3 — a technique where a model is trained on the outputs of another. As Business Insider reported, Kratsios alleged that Moonshot developed a sophisticated internal platform to conduct large-scale distillation, switching between multiple methods of access to avoid detection.

While distillation itself is a common and legitimate practice in machine learning, Kratsios argued that “large-scale, covert industrial distillation aimed at stealing proprietary U.S. technology” crossed a line. Beijing has dismissed the claims as “groundless.”

Some experts question whether the timeline supports the accusation — Anthropic’s Fable was released on July 1, and Kimi K3 launched just 15 days later on July 16, a window many consider too short to distill and train a 2.8-trillion-parameter model.

Sanctions on the Table

US Treasury Secretary Scott Bessent has warned that further sanctions could be coming. According to TechCrunch, Bessent posted on X: “Open source is not open season on American IP. When [Chinese] firms conduct covert, industrial-scale distillation attacks that cross the line into IP theft, sanctions and Entity List designations will be on the table.”

The threat comes amid a broader debate in Washington over how to respond to the influx of Chinese open-source models. Some argue for restrictions to preserve America’s technological edge, while others point to a paradoxical effect of US policy.

The Policy Paradox

Export controls imposed by the Trump administration in June 2026 on Anthropic’s Fable and Mythos models — which kept them offline for more than two weeks — inadvertently created an opening for Chinese competitors. Z.ai’s GLM-5.2 launched during that window.

“Restricting an American model can immediately create an opening for a Chinese competitor,” said Anastasios Angelopoulos, co-founder and CEO of Arena, an AI evaluation platform.

Adding to the complexity, major US tech firms including Microsoft, Meta, and Nvidia signed an open letter on July 24 backing open AI models — a stance that sits uneasily with potential restrictions on Chinese open-source offerings.

Goldman Sachs concluded in a July research report that Chinese AI models are reaching a “critical stage” for widespread adoption, particularly as surging “agentic” AI usage — where models autonomously conduct multi-step tasks — compounds cost differences through token usage scaling.

What Lies Ahead

Chinese model vendors are expected to leverage their open-source approach to promote global adoption. At the World AI Conference in Shanghai on July 17, President Xi Jinping championed open-source AI and pledged Chinese involvement in raising AI capabilities in developing nations.

Yet significant challenges remain. While Chinese models rival US systems on many benchmarks, they still lag in overall full-range capabilities. And the economics are punishing — Z.ai reported revenue of $107 million in 2024 alongside a net loss of $694 million, highlighting the sustainability questions facing the sector.

As Morningstar analyst Chelsey Tam noted: “Both China and the U.S. will want to encourage widespread adoption of their AI ecosystems, while safeguarding technologies that could materially strengthen strategic rivals.”

For now, the momentum is unmistakable. Krikorian’s advice to developers is straightforward: “I would highly recommend anyone doing any serious AI load to at least evaluate it.” As Chinese AI models continue to improve and US policies evolve, the competition for the future of artificial intelligence is only just beginning.