Alibaba unveiled a new state-of-the-art AI model: Qwen4-Max. The company claims that its performance is comparable to that of models from Anthropic and OpenAI. This would mark a new Chinese challenge to the American market leaders.
Bloomberg reports this, based on Alibaba’s announcement. The e-commerce giant, which has now also turned its focus to AI, said that Qwen4-Max ranked very highly in coding benchmarks. It also reportedly outperforms both Anthropic’s Claude Fable 5.1 and the Kimi K3.1 Pro from domestic Chinese competitor Moonshot. Alibaba has not made the model open source, which would allow users to download and customize it, but said it will soon make smaller versions of the series available.
Chinese companies, including Alibaba, Moonshot, and DeepSeek, have long been challenging U.S. dominance in AI by developing a series of new models. The new models are making particular strides in the areas of coding and reasoning. The launch of Qwen4-Max joins a wave of recent Chinese releases, including Tencent’s Hunyuan 4 and Moonshot’s Kimi K3.1 Pro.
According to Bloomberg, Alibaba’s Qwen series is the most-downloaded open-weight model family in the world to date. Following an ambitious technological push led by CEO Eddie Wu, Alibaba has managed to attract a large user base. The news agency notes that the company is building a complete AI ecosystem, ranging from AI models to its own AI chips.
Investments in data centers
Eddie Wu said in September that T-Head, Alibaba’s chip division, would introduce China’s most powerful AI chips to the domestic market. T-Head recently filed for an initial public offering in Hong Kong. The Hangzhou-based company has set the achievement of artificial general intelligence (AGI) as its goal. It is also rapidly expanding its computing power, with Wu intending to build 20 gigawatts of data center capacity worldwide by 2032.
Alibaba has introduced enhanced security measures in Qwen4-Max, following previous incidents that raised concerns about the cybersecurity risks of open-weight models. This follows similar steps taken by Tencent and Moonshot.