China’s AI boom has just picked up speed as days after startup Moonshot AI rocked the tech world with its Kimi K3 system, e-commerce giant Alibaba launched a preview of its latest flagship model, Qwen3.8 Max. The tech giant positioned the powerful new model as directly comparable to top global frontier systems, describing its performance as second only to Anthropic’s industry-leading Fable 5. According to a report by Bloomberg, the back-to-back releases from Chinese companies signal that they are rapidly ‘erasing’ the capabilities gap with Silicon Valley.
What is Alibaba’s 2.4 trillion-parameter AI model
Alibaba rolled out preview access to Qwen3.8 Max, allowing software developers to immediately test the system through its coding platforms, including Qoder. While full technical specifications remain under wraps, Alibaba reportedly confirmed that Qwen3.8 Max boasts a massive 2.4 trillion parameters, which is the key metric measuring an AI’s internal scale and processing capability.The company also announced plans to release the model as an “open-weight” system soon, allowing outside developers to download and customise the code freely.
Win-win for Alibaba
The product rollout tops off a streak of high-profile wins for the e-commerce giant. Chinese regulators recently approved Apple Intelligence in the country, selecting Alibaba as Apple’s primary technology partner to power local AI features for iPhone users across China. Alibaba’s release comes right on the heels of Moonshot AI’s announcement of Kimi K3, a massive system containing 2.8 trillion parameters.Backed by tech heavyweights Alibaba and Tencent, Moonshot hailed Kimi K3 as its “most capable flagship model to date.” The startup confirmed that full access to the system, including its advanced coding, reasoning, and automated knowledge-work features, will officially open to the public on July 27.Upon its release, Kimi K3 will become the world’s first open-source model in the 3-trillion-parameter class, allowing global researchers and businesses to run top-tier AI locally without relying on US cloud infrastructure.
