Alibaba Officially Releases Multi-Modal Smart Agent Model Qwen3.7-Plus, Opening Commercial API Access

According to WatchTower Beating monitoring, the Alibaba Tongyi Q&A team has officially released the new multimodal large model Qwen3.7-Plus, positioned as a multimodal intelligent agent base. Qwen3.7-Plus has now landed on the Alibaba Cloud Hundred Refinements Model Studio platform and opened up API commercial access.

On May 19, the Q&A team quietly launched the Qwen3.7 preview version Preview on the official website for testing. This official release marks the full landing of the Qwen3.7-Plus official version and commercial interface. Unlike Qwen3.7-Max, which serves as the flagship deep inference model, Qwen3.7-Plus is a mixed intelligent agent model focused on multimodal interaction. Qwen3.7-Plus is not open source but is provided as a proprietary closed-source model for API services.

Qwen3.7-Plus is able to unify graphical user interface (GUI) and command-line interface (CLI) operations in a single closed loop, while supporting visual perception, screen reading, and terminal code execution. By integrating visual and language capabilities, Qwen3.7-Plus can directly infer and generate executable SVG code or front-end pages based on user-uploaded web screenshots, videos, or design prototypes. In terms of technical architecture, Qwen3.7-Plus has undergone deep optimization for multimodal intelligent agents in complex long-range tasks, supporting seamless generalization across various mainstream intelligent agent frameworks, further enhancing the accuracy of perception, inference, and retrieval for enhanced question-answering.