Alibaba Cloud launches Qwen3.6-Plus with upgraded multimodal & agentic coding capabilities

Alibaba Cloud has released Qwen3.6-Plus, establishing a new standard in agent-driven coding and multimodal artificial intelligence. Building on the earlier Qwen3.5 release, Qwen3.6-Plus delivers upgrades in agentic coding, enabling more accurate code generation and advanced reasoning for both general and specialized tasks.

This model introduces comprehensive improvements across coding agents, general AI agents, and tool usage. These advances are achieved by tightly integrating reasoning, memory, and execution capabilities, resulting in reliable performance for a variety of engineering workflows. For software developers, Qwen3.6-Plus closely matches major industry benchmarks in code repair, while also excelling in complex terminal operations and automated task execution.

Following these enhancements in coding, Qwen3.6-Plus demonstrates leadership in long-horizon planning and tool-calling within general-purpose agent scenarios. The model achieves new records in evaluations ranging from challenging science, technology, engineering, and math (STEM) reasoning to precise information extraction from ultra-long contexts and strong adaptation to multilingual environments.

In addition, Qwen3.6-Plus advances its multimodal capabilities, enhancing both reasoning and task execution with broad applicability. The solution is now generally available via Alibaba Cloud Model Studio’s API, with seamless integration support for third-party coding assistants such as OpenClaw, Claude Code, Qwen Code, OpenCode, Kilo Code, and Cline.