Qwen 3.6 Plus Uncensored

• - 🏢 Alibaba's flagship Qwen3-family agentic reasoning model, released April 2026.
• - 📏 1 million token context window enabled by default.
• - 👁️ Multimodal input: processes both text and images.
• - 🔧 Native function calling, tool use, and web search.
• - 🧠 Mixed/always-on reasoning with preserved thinking across steps.
• - 🎯 Built for repository-level agentic coding workflows.

Alibaba

Alibaba Group is a Chinese multinational technology company founded in 1999 and headquartered in Hangzhou, Zhejiang. Originally built around e-commerce and cloud computing, Alibaba has become one of the most prolific contributors to open-weight AI research,…

Qwen 3.6 Plus is positioned in the "Plus" tier of Alibaba's Qwen3 lineup, sitting below the Max tier exemplified by Qwen 3.7 Max. It is designed explicitly for "real world agents," integrating deep reasoning, long-context memory, and tool execution into a single perceive-reason-act loop. Its headline feature is a 1-million-token context window by default, which enables full-codebase reasoning without retrieval workarounds.

According to Alibaba, Qwen 3.6-Plus is a significant step up from the prior 3.5-Plus generation, directly addressing community feedback to deliver a more stable foundation for agentic coding, sharper multimodal perception, and stronger long-context extraction. A "thinking preservation" capability retains reasoning context across conversation history and agent steps, reducing overhead in iterative development.

Within the broader Qwen3.6 release, Alibaba also shipped sibling variants such as Qwen 3.6 35B A3B FP8 and the Qwen3.6 35B A3B Uncensored, alongside the smaller Qwen 3.6 27B. These span a range of sizes and deployment options across the family.