Wukong Integrates Qwen3.6-Plus, Alibaba's Most Powerful Programming Model

ME News reports that on April 2 (UTC+8), Alibaba officially launched its next-generation large language model, Qwen3.6-Plus. The company’s flagship enterprise AI application, Wukong, was the first to integrate the new model. Qwen3.6-Plus delivers significant performance improvements across code generation, agents, reasoning, and native multimodal capabilities. In authoritative benchmarks such as the SWE-bench series for agent programming and Claw-Eval for real-world agent tasks, Qwen3.6 outperforms models with 2 to 3 times its parameter size—including GLM-5 and Kimi-K2.5—establishing itself as the most powerful domestic model for programming, nearing the performance of global leaders like the Claude series. For Wukong users, the integration of this new model brings three key enhancements: First, agent programming capabilities are dramatically improved compared to general-purpose models; AI transitions from a “co-pilot” to a true “collaborator” capable of independently handling subtasks—writing cross-file code, running tests, and iterating fixes autonomously. Second, long-range task planning and execution capabilities are significantly strengthened, enabling more reliable decomposition of complex, multi-step workflows common in enterprise environments, efficient resource allocation, and consistent delivery of results. Third, the model offers outstanding cost efficiency—with input costs as low as RMB 2 per million tokens—greatly lowering the barrier to enterprise-scale adoption. As Alibaba’s flagship enterprise AI application, Wukong plays a critical role in bridging cutting-edge model advancements with real-world business needs. By being the first to integrate Qwen3.6-Plus, Wukong delivers an enterprise-grade agent experience with superior programming capability, enhanced execution quality, and unmatched cost efficiency. For example, in solo development scenarios, users can describe requirements in natural language, and Wukong autonomously completes the entire workflow—from solution breakdown and code generation to website creation and testing validation. In knowledge-intensive roles such as solo legal work, Wukong can extract key information across documents, compare clause changes, identify risks, and generate actionable recommendations. In business process automation scenarios like finance or manufacturing, employees simply state their goal, and Wukong autonomously performs cross-system queries, form submissions, and approvals—transforming cumbersome workflows into “one-sentence” intelligent services. Wukong’s early integration of Qwen3.6-Plus represents a pivotal milestone in Alibaba’s strategy of aligning AI models with practical applications. Wukong’s product lead stated: “Competition in enterprise AI ultimately comes down to the product of model capability × domain understanding. The breakthroughs in agent programming and long-range task planning within Qwen3.6-Plus provide Wukong with a new foundation to solve complex enterprise pain points. Meanwhile, DingTalk’s years of deep expertise in enterprise markets give Wukong the contextual insight needed to precisely embed these capabilities into real-world workflows.” Starting today, Wukong premium users on version v0.9.20 and above can access the Qwen3.6-Plus model ahead of others. (Source: Ifnar)