Alibaba is pushing forward: just one day after releasing Qwen 3.5 Omni, they’re dropping Qwen 3.6-Plus — a new flagship model specifically built for agentic coding and enterprise applications.
What can Qwen 3.6-Plus do?
The specs are impressive. One million token context natively — no extra feature flag, no beta. On agentic coding benchmarks, the model reportedly lands in the neighborhood of Claude 4.5 Opus. It can break down complex programming tasks into steps, write code, test it, and debug iteratively.
As a native multimodal model, it also generates frontend code from screenshots and design drafts. Useful if you want to go from mockup to prototype fast.
Availability and pricing
Qwen 3.6-Plus is available through Alibaba Cloud’s Model Studio and Qwen Chat. The price: roughly $0.29 per million input tokens. It’s also compatible with third-party tools like OpenClaw, Claude Code, and Cline.
Alibaba’s bigger play
The model is being integrated into Alibaba’s own ecosystem — including Wukong, an AI-native enterprise platform for multi-agent workflows, and the Qwen App, Alibaba’s consumer AI application.
What it means
Alibaba means business. Two models in two days, plus an aggressive pricing strategy. In the LLM race, Qwen is increasingly positioning itself as a real alternative to Western models — especially for developers and companies watching their budgets. One million tokens of context at a fraction of the competitor’s price? That’s going to get attention.
Sources: