Alibaba Advances AI Capabilities With New Models, Chips
- tech360.tv
- 8 hours ago
- 2 min read
Alibaba Group Holding has unveiled a comprehensive suite of new artificial intelligence products, including advanced models, cloud infrastructure, and chips. The Chinese tech giant aims to position itself as China's "AI factory" as autonomous agents become more prominent.
Liu Weiguang, senior vice-president of Alibaba's cloud computing unit, described AI as a new form of manufacturing that generates revenue through "training and inference factories." He stated that Alibaba is constructing China's AI factory, describing it as the only AI and cloud company in the country operating all five layers of the full AI stack. This stack encompasses chips, agentic cloud, AI models, model service platforms, and agentic applications.
The company announced Qwen3.7-Max, its latest proprietary large language model, designed as a robust foundation for AI agents. This model reportedly excels at agentic coding, complex reasoning, and long-horizon tasks, which are multi-step missions requiring continuous decision-making. Alibaba stated Qwen3.7-Max can operate autonomously for up to 35 hours without performance degradation.
Zhou Jingren, chief AI architect of the newly formed Alibaba Group Technology Committee, indicated the model consistently ranks among the top tier on various benchmarks. Jingren added that it outperformed all other AI models in China, although it still trails leading US models from companies such as Anthropic, Google, and OpenAI in text and vision capabilities.
Alibaba did not mention open-sourcing smaller versions of its new flagship model. This aligns with a broader trend in the Chinese industry moving away from fully open-sourcing powerful models amid increasing commercialisation pressure. The company kept its Qwen3.6-Max and Qwen3.6-Plus models proprietary, but open-sourced Qwen3.6-27B, a smaller model described as "dense" and possessing flagship-level coding power.
Alibaba also introduced the Zhenwu M890, an AI training and inference chip. The company stated this hardware delivers three times the performance of its predecessor, the Zhenwu 810E, making it exceptionally well-suited for complex agentic AI workloads. Alibaba's chip design arm, T-Head, previously noted that the Zhenwu 810E performed on par with Nvidia's H20, a GPU tailored for the Chinese market.
Gao Hui, T-Head vice-president, indicated that Alibaba had shipped 560,000 Zhenwu chips to more than 400 clients across 20 industries, including carmakers and financial services. T-Head plans to launch the Zhenwu V900 processor in the third quarter of 2027, followed by the Zhenwu J900 in the third quarter of 2028. Alibaba further launched the Panjiu AL128 Supernode Server, equipped with the new Zhenwu M890 chip. First debuted with an older chip, the server infrastructure has dramatically improved the handling of large-scale concurrent requests from agents.
Alibaba launched a new suite of AI products, including models, cloud infrastructure, and chips.
The company aims to be China's "AI factory," claiming to be the only AI and cloud company operating all five layers of the AI stack.
Its new Qwen3.7-Max large language model reportedly excels in agentic coding and complex reasoning.
Source: SCMP
