【AI+9988】Alibaba releases programming model Qwen3.6-Plus, emphasizing intelligent agent programming capabilities

Alibaba （09988） On Thursday (the 2nd), Alibaba officially launched Qwen3.6-Plus, its new-generation large language model. The model represents a significant upgrade in its technical architecture, with a particular focus on optimizing “agentic coding” and native multimodal reasoning capabilities, aiming to meet the growing demand from enterprise-level development for automated navigation and more complex engineering handling.

According to a technical white paper released by Alibaba Cloud, Qwen3.6-Plus has shown notable improvements over its predecessor models in execution efficiency and reasoning stability. Its core specifications include:

A 1-million-Token context window: The model comes standard with support for 1 million tokens of context input (Token broadly refers to AI computing resource measurement or service billing units). This allows developers to import an entire code repository (Repository-level) or long-form technical documentation in a single step, addressing the fragmentation problem where previous models had to frequently switch contexts when handling large projects.

Optimized Mixture of Experts (MoE) architecture: It adopts a new generation of mixture architecture, improving the model’s response speed during complex logical reasoning. Third-party test data shows that its generation speed is faster than that of comparable products in specific benchmark tests.

A “Perceive-Reason-Execute” feedback loop: Unlike traditional one-way generation, Qwen3.6-Plus has been deeply optimized for a “capability loop,” enabling it to complete a closed loop within a single workflow—from goal decomposition, to code writing, to self-testing.

Strengthened multimodal and front-end development capabilities: In multimodal application scenarios, Qwen3.6-Plus demonstrates direct transformation capability from visual inputs to code outputs. The model can now interpret UI screenshots, hand-drawn sketches, and product prototype diagrams, and can generate functional front-end code in real time. This technological breakthrough is viewed as a key milestone to shorten product development cycles.

Enterprise AI application “Wukong” has already been connected

On the day of release, Alibaba’s enterprise AI application “Wukong” has already completed model integration. Wukong member users can try Qwen3.6-Plus in version v0.9.20 and above. Developers can call the model’s API service via Alibaba Cloud’s Bailian platform.

According to the latest pricing on Alibaba Cloud’s Bailian platform, the lowest price for Qwen3.6-Plus is RMB 2 yuan per one million Tokens input; Qianwen 3.6 has also been launched on Wukong and other Alibaba AI applications and platforms such as the Qianwen app. Alibaba also announced that it will roll out the Qwen3.6-Max version with an even larger parameter scale.

		Finance Hot Talk
	





	Will the Iran-Israel conflict turn into a prolonged war? Has the market underestimated the risk of a global economic recession?