Alibaba Group has released a new version of its large language model, Qwen3.5, as competition in China’s AI sector increasingly centers on so-called AI agents.
The company launched the model late Monday, just ahead of the Lunar New Year, offering two versions. One allows developers to download and run the model on their own systems, while the other runs through Alibaba’s cloud platform. Alibaba said the update improves both performance and operating costs compared with earlier releases.
Qwen3.5 also reflects a broader shift in how companies build AI. The model supports agentic functions, meaning it can carry out tasks across multiple steps with limited supervision. It also works with open-source agent frameworks that have gained traction in recent weeks.
Interest in AI agents has surged since Anthropic released new tools that sparked concern across parts of the software market. In China, rivals such as ByteDance and Zhipu AI have also rolled out updates focused on similar capabilities.
Marc Einstein of Counterpoint Research said companies are preparing for a future where AI agents disrupt existing internet business models. Those that move too slowly, he warned, could struggle to adapt.
Alibaba claims Qwen3.5 performs on par with leading models from OpenAI and Google DeepMind, though the benchmarks are self-reported. The model also supports more than 200 languages, a sharp increase from earlier versions.
Alibaba said it plans to release additional open-weight models in the coming weeks, signaling it does not intend to slow down.
Related Readings:
