Alibaba Unveils Qwen3.8, a 2.4 Trillion Parameter Large Language Model
On August 3rd, Alibaba officially launched its next-generation foundational large language model, Qwen3.8. This new model boasts a total of 2.4 trillion parameters, representing a significant advancement over previous versions. Alibaba has reported substantial improvements in Qwen3.8's capabilities, particularly in the areas of programming and professional office tasks. In a third-party ranking released today by Arena, Alibaba's Qwen model was positioned just behind Anthropic's Claude series, indicating strong competitive performance. The API for Qwen3.8 is now available on the Qianwen AI platform. Furthermore, Qwen3.8 has been integrated into 'Qianwen Office,' an Agent product that Alibaba also launched concurrently. Looking ahead, Alibaba plans to open-source the Qwen3.8-Max model next week, along with the Qwen3.8-27B model.
Alibaba's release of Qwen3.8, a 2.4 trillion parameter model, signifies a continued escalation in the scale and capability of foundational AI models. The emphasis on enhanced coding and professional office functionalities suggests a strategic focus on practical enterprise applications, moving beyond general-purpose AI. Its competitive placement on the Arena leaderboard indicates the growing maturity of Chinese AI development in the global landscape. The planned open-sourcing of Qwen3.8-Max and Qwen3.8-27B aligns with broader industry trends that balance proprietary development with community contribution, potentially accelerating innovation and adoption through wider access. This move could foster a more diverse ecosystem of AI applications built upon Alibaba's architecture, while also providing valuable data points for understanding the performance and limitations of extremely large models in real-world scenarios.
AI-generated to prompt reflection — not editorial opinion, not advice, not a statement of fact. How this works.
