Alibaba Unveils Qwen3.7-Plus, Claiming Top China Ranking in Multimodal AI Agents
Alibaba has released Qwen3.7-Plus, a multimodal upgrade to its Qwen3.7 foundation model, achieving China's highest ranking in Vision Arena benchmarks while delivering advanced visual reasoning and cross-modal task execution capabilities for enterprise automation.
Alibaba has officially launched Qwen3.7-Plus, a significant evolution of its Qwen3.7 large language model designed specifically as a multimodal interactive hybrid AI agent. Announced on June 2, the release positions the model as a unified vision-language foundation capable of executing complex tasks across graphical interfaces, command-line environments, and software workflows.
The Qwen3.7-Plus model retains core competencies in text generation, code synthesis, tool utilization, and productivity automation while substantially enhancing visual understanding and cross-modal reasoning. It accepts diverse input types including static images, video streams, screen captures, web content, and textual data, enabling autonomous operation within GUI and CLI environments to manage sophisticated office and software processes.
Performance metrics place Alibaba at the forefront of Chinese AI development. According to Vision Arena evaluations, Qwen3.7-Plus has secured Alibaba's position within the global top five while achieving the top domestic ranking. In pure text assessments, the model approaches the capabilities of flagship Max-tier systems, demonstrating particular strength in coding agents, general-purpose reasoning, instruction following, and multilingual tasks.
Multimodal benchmark results indicate significant advancement in visual reasoning chains and tool-augmented execution. The model shows substantial improvements on specialized evaluations including BabyVision, MathVision, ScreenSpot Pro, OSWorld-Verified, and AndroidWorld—tests measuring screen comprehension, mathematical visual reasoning, and mobile operating system navigation.
Enterprise users and developers can access Qwen3.7-Plus immediately through Alibaba Cloud's Bailian service platform or via the Qwen Studio interface. The release represents Alibaba's strategic push toward agentic AI systems capable of autonomous software interaction and comprehensive workflow automation through unified multimodal perception.