China's AI competition with the United States has intensified further with Alibaba unveiling a preview of its latest artificial intelligence model, Qwen 3.8. The company claims the model is among the most powerful AI systems currently available, second only to Anthropic's Claude Fable 5. The announcement was made at the World AI Conference (WAIC) in Shanghai, highlighting the rapid progress Chinese AI companies are making in narrowing the gap with their US counterparts.
According to Alibaba, Qwen 3.8 features 2.4 trillion parameters—a key measure of an AI model's scale and complexity—and delivers significant improvements over its predecessor, Qwen 3.7 Max. The company said the model performs better across coding, reasoning and complex tasks such as full-stack software development, data analysis and office productivity workflows.
A preview version of Qwen 3.8 is currently available through Alibaba's Token Plan subscription service as well as its Qoder and QoderWork platforms. The company also announced that the model will soon be released as an open-weight AI system, allowing developers to access its trained parameters while keeping the underlying training data and source code proprietary.
A RIVALRY TAKING SHAPE IN CHINA'S AI INDUSTRY
Alibaba's announcement appears to coincide with growing momentum around Moonshot AI, a Beijing-based artificial intelligence company that introduced its Kimi K3 model just days earlier.
Kimi K3 is even larger, with 2.8 trillion parameters, making it the world's largest open-source AI model currently available.
Moonshot has claimed that Kimi K3 delivers frontier-level performance across several benchmark tests and outperformed Anthropic's Claude Fable 5 and OpenAI's GPT-5.6 Sol in evaluations measuring software engineering capabilities, including building software projects from scratch and completing long-duration coding tasks.
However, those performance claims are based on Moonshot's internal testing and have not been independently verified by external benchmarking organisations.
Similarly, Alibaba's assertion that Qwen 3.8 outperforms Kimi K3 has also not yet been validated through independent evaluations.
Nevertheless, both product launches reflect a broader trend within China's AI industry. Leading Chinese AI laboratories have now entered the multi-trillion-parameter era and are increasingly benchmarking their models directly against those developed by Anthropic, OpenAI and Google.
The announcements also carried broader political significance. Chinese President Xi Jinping made his first in-person appearance at this year's World AI Conference and reaffirmed China's commitment to "encouraging open source, openness, collaboration and sharing."
He added that such an approach would help ensure that "all sectors and businesses can benefit from AI."
