Alibaba pushes AI boundaries with upgraded Qwen VLo visual engine

Alibaba’s upgraded Qwen VLo model enables both text-to-image and image-to-image creation, featuring a “progressive generation” function that lets users watch images render in real time, enhancing transparency and control in AI-assisted visual content creation

Alibaba Group has introduced an advanced version of its visual AI model, Qwen VLo, expanding its suite of generative tools designed for text- and image-based interaction. The release marks a significant step in Alibaba’s AI strategy, as the Chinese tech giant moves deeper into the development of multimodal systems capable of processing and generating across various media formats.

The upgraded Qwen VLo supports both text-to-image and image-to-image capabilities, allowing users to describe an image for generation or upload a visual for AI-assisted editing using written instructions. One of its standout features is “progressive generation,” which offers users a real-time view of how an image is constructed step by step—a tool aimed at enhancing transparency and user control in content creation.

With commands like “Generate a serene mountain landscape” or “Add a red balloon in the background,” users can prompt the AI to deliver tailored results, making the tool suitable for content creators, designers, and app developers.

Alibaba sharpens focus on AGI

The launch is part of Alibaba’s broader AI roadmap. CEO Eddie Wu recently reaffirmed the company's focus on artificial general intelligence (AGI), and the Qwen brand has become central to this goal. Beyond visual intelligence, Alibaba is also building models that support voice, video, and text interaction—all optimized for mobile performance.

Qwen VLo also plays a role in powering Alibaba’s updated Quark assistant, signalling the company’s intent to bring enterprise-grade AI capabilities to mainstream consumer applications.