Alibaba released Qwen3.6-Plus on April 2, the latest flagship in its Qwen large language model series and the company's third proprietary model launch in a single week following the Qwen3.5-Omni-Plus rollout.

What's New

The model ships with a 1 million-token context window by default and focuses heavily on agentic coding - the ability to autonomously plan, write, test, and iterate on code across entire repositories. According to Alibaba's benchmarks, Qwen3.6-Plus matches Claude Opus 4.5 on SWE-bench, the standard test for real-world code repair tasks.

Beyond coding, the model adds improved multimodal reasoning across images, documents, and video. Alibaba says it handles complex terminal operations, automated task execution, and long-horizon planning tasks, positioning it as an "all-rounder" for autonomous workflows.

Pricing and Availability

Qwen3.6-Plus is available immediately via Alibaba Cloud's Model Studio API at roughly $0.29 per million input tokens and $1.71 per million output tokens - significantly cheaper than comparable Western models. It works with third-party coding tools including Claude Code, Cline, and OpenCode.

Bigger Picture

The rapid-fire release schedule signals Alibaba's aggressive push to compete with Anthropic and OpenAI in the agentic AI space. The model will integrate into Wukong, Alibaba's enterprise AI platform currently in invitation-only beta, which connects to DingTalk's 20 million-plus user base. Alibaba has also confirmed that selected models from the Qwen3.6 series will be open-sourced, continuing the company's dual commercial and open-source strategy.