Chinese AI firms launch Kimi K3 and Qwen 3.8, challenging US dominance
Juste les faits

Chinese AI firms launch Kimi K3 and Qwen 3.8, challenging US dominance

Résumé

Beijing-based Moonshot AI and Alibaba introduced new large language models that claim performance near top US systems and are priced lower, prompting market and policy reactions in the United States.

Beijing-based startup Moonshot AI unveiled its flagship model Kimi K3 on Friday, saying it trails only OpenAI’s GPT-5.6 Sol and Anthropic’s Claude Fable 5 in performance. The company priced the model at $15 per million output tokens, roughly half the cost of comparable US offerings, and reported that demand temporarily forced a pause on new subscriptions. A few days later, Alibaba previewed its Qwen 3.8 model, describing it as “one of the most powerful models available today” and ranking it second only to Fable 5.

Both firms said the models will be released as open-weight, allowing developers to download, use and modify the underlying parameters, a contrast to the closed-source approach of most leading US AI labs. Analysts note that Chinese models have been narrowing the performance gap with US systems and are often cheaper to run, leading some US companies to consider them as cost-effective alternatives.

The launches have sparked reactions in financial markets and among policymakers, who have referenced the competitive implications for US AI investment and national security. Some observers warn that wider availability of capable Chinese models could affect the economics of US data-center and chip spending, as well as the strategic calculus around AI access and cybersecurity.

While the models have not been fully released for independent testing, the claims highlight a growing trend of Chinese AI firms producing systems that could rival top US offerings, suggesting a shift from occasional surprises to a more regular competitive dynamic.

Continuez à lire — gratuitement

Créez un compte gratuit pour suivre l'actualité. Aucune carte requise.