Alibaba's Qwen team launches Qwen3.7-Plus with vision and agentic features

Alibaba's Qwen team released Qwen3.7-Plus, a new multimodal large language model available through Alibaba Cloud's Bailian platform. The model understands image and video inputs and layers on agentic capabilities: deep reasoning, self-programming, tool invocation and autonomous iteration.

The upgrade's core additions over the previous Qwen3.7 generation are multimodal understanding and the agentic feature set — moving Qwen from a strong text model toward an action-taking system that can perceive, reason, call tools and iterate on its own outputs.

Strategically, Qwen3.7-Plus keeps Alibaba competitive in the agentic-AI race and anchors its Bailian cloud platform as the distribution channel for Chinese and global developers. Qwen has been one of the most influential open and semi-open model families, and continued rapid iteration matters for Alibaba's cloud ambitions against domestic rivals like DeepSeek and international frontier labs.

The caveats: 'Plus' is a hosted Bailian offering rather than a fully open release, and self-reported agentic capabilities need independent evaluation to gauge how Qwen3.7-Plus stacks up against GPT-5.5, Claude and Gemini on real tool-use reliability. Availability and pricing outside China are also open questions for international developers weighing it.