Jakarta, ThedailyID — Chinese technology giant Alibaba has launched Qwen 3.8-Max, its latest artificial intelligence model, for global users. The company says the model performs close to Anthropic’s Claude Fable 5, one of the world’s most advanced AI systems.
The release also marks Alibaba’s return to open-source development for its flagship AI models. Earlier this year, the company kept several of its most advanced models proprietary.
The launch comes as Chinese AI companies race to narrow the technology gap with leading U.S. developers through increasingly capable foundation models.
According to the South China Morning Post, Qwen 3.8-Max features 2.4 trillion parameters and supports a one-million-token context window. That allows the model to process roughly 750,000 words in a single prompt.
Developers can now access the model through Alibaba Cloud’s Model Studio API. It is also available on QwenWork, Alibaba’s AI-powered workplace platform.
Following the announcement, Alibaba’s Hong Kong-listed shares climbed 7 percent, closing at HK$125.20.
Alibaba said Qwen 3.8-Max can process multiple data types, including long documents, television series, and live streams, to build searchable knowledge bases.
The company also claims the model can redesign software from screenshots, create interactive games and educational animations, and convert two-dimensional floor plans into 3D visualizations.
Alibaba designed the model for advanced coding, complex research, and long-running AI workflows.
According to company benchmarks, Qwen 3.8-Max ranks fifth on Text Arena and second on Vision Arena, two AI leaderboards based on user voting. It trails only a handful of Anthropic’s Claude models.
Technically, Qwen 3.8-Max uses a sparse mixture-of-experts architecture with hybrid attention. Although it contains 2.4 trillion parameters, it activates only about 95 billion, or roughly 4 percent, during inference. Alibaba says this approach improves computing efficiency while maintaining high performance.
The launch comes amid intensifying competition among Chinese AI developers. Last month, Moonshot AI introduced Kimi K3, a 2.8-trillion-parameter multimodal model with a one-million-token context window. DeepSeek has also upgraded its V4-Flash API and begun testing new software for managing AI agent workflows.
Alibaba also said its newly launched QwenWork platform can draft documents, analyze data, generate audio and video, and build websites. The company plans to integrate the platform into its workplace application DingTalk.
