China has emerged as a significant player in the U.S. artificial intelligence market. Mozilla’s Chief Technology Officer, Raffi Krikorian, has recently started using the Kimi K3 model from Chinese AI startup Moonshot. This switch came shortly after the model’s launch, citing its responsive nature compared to other options like Anthropic’s Claude Fable chatbot.
Chinese AI models such as Z.ai’s GLM-5.2 and Moonshot’s Kimi K3 are gaining recognition for being cost-effective and efficient, attracting attention from companies like Coinbase. These models are seen as viable alternatives to more expensive American systems, despite growing concerns from U.S. tech giants.
The U.S. has imposed several restrictions on China’s access to advanced technology, including AI chips. These measures have sparked debates, with accusations of illicit technology extraction from companies like Anthropic. China denies such claims, asserting its models are developed independently.
Chinese AI models are cheaper and good enough for most users.
The rapid evolution of Chinese AI technology began with DeepSeek’s model last year, which was competitive with U.S. counterparts yet more affordable. New models from Z.ai and Moonshot this year, along with Alibaba’s Qwen3.8 Max AI, have kept China on the forefront of AI development.
With high demand for cost-effective AI solutions, Chinese models are reaching a pivotal point in their adoption. Goldman Sachs reports an increasing need for AI models that manage complex tasks efficiently. Factors like token pricing amplify the cost differences, drawing users like Curt Meinhold to opt for these solutions over pricier alternatives.
Moonshot’s Kimi K3 is popular, but Chinese AI models have limitations.
Moonshot’s Kimi K3 model gained significant popularity on platforms such as OpenRouter. However, despite their affordable edge, Chinese models still lag behind in some capabilities compared to U.S. leaders. Yet U.S. firms are also seeking ways to reduce costs, which might inadvertently boost Chinese models.
Most Chinese AI models offer open-source access, contrasting the closed-source approach of U.S. firms like Anthropic. Experts indicate that Chinese open-source models are nearing the effectiveness of their closed American counterparts. Such developments are fostering the adoption of open AI models globally.
Chinese AI models look to expand globally.
Driven by competition, Chinese AI startups are expanding globally, fueled by strong state support and strategic international initiatives. Efforts include embedding AI into devices and raising global AI capabilities, especially in developing nations. With increasing funding and public share offerings, Chinese enterprises are furthering their overseas influence.
Both China and the U.S. are likely to promote their AI technologies while protecting them from strategic competitors. Chelsey Tam from Morningstar and others note this competitive landscape is continually evolving, with intra-China competition equally intense. Despite strong growth, sustainability concerns persist among these startups due to high investment costs.
Overall, Chinese AI models are poised to advance further, including in the U.S., according to Krikorian. Thorough evaluation of these models is recommended for serious AI applications.
Reported by O’Brien from Providence, Rhode Island. Contributions from Kelvin Chan in Toronto.