Alibaba (09988) releases the strongest domestic programming model Qwen3.6-Plus

On April 2, Alibaba (09988) released the new generation large language model Qwen3.6-Plus. Qwen3.6 has native multi-modal understanding and reasoning capabilities, significantly enhancing overall performance. In authoritative evaluations such as the SWE-bench series and Claw-Eval for real-world intelligent agent tasks, Qwen3.6's programming performance surpasses models with 2 to 3 times the parameters such as GLM-5 and Kimi-K2.5, becoming the strongest domestic model in programming ability and approaching the world's strongest programming model, the Claude series. In real test scenarios such as front-end web development and warehouse-level complex tasks, Qwen3.6 can autonomously decompose tasks, plan paths, test modifications until the task is completed, demonstrating a new breakthrough in multi-modal intelligent agent programming (Agentic Coding) and making "ambient programming" where AI writes code driven by a sentence truly usable. Currently, Qwen3.6-Plus has been launched on Alibaba Cloud Hundred Refinement, with a minimum price of 2 yuan per million Tokens input; Qwen3.6 has also landed on Alibaba AI applications and platforms such as Wukong and Qianwen APP, allowing developers and individual users to experience the latest models firsthand. Public information shows that Qwen3.5 was released in February 2026, and in less than a month and a half, Qwen has evolved to Qwen3.6, with iteration noticeably accelerated. It is reported that the Qwen3.6 series will open source other size models, and the flagship model Qwen3.6-Max with even stronger performance will also be released soon.