At the 2026 World Artificial Intelligence Conference (WAIC), Sensetime-W Chairman and CEO Xu Li unveiled the company's flagship image creation model, the SenseNova-U1 Pro.
This model is designed to conduct deep creation through an inherent interleaving of text and image reasoning, delivering results directly.
As previously reported, the U1 Pro project is led by Sensetime-W co-founder and Chief Scientist Lin Dahua.
It aims to compete with OpenAI's GPT-Image 2, with a focus on developing an image generation model capable of 'thinking'.
On a technical level, the model is built upon a multimodal agent foundation that natively unifies 'understanding, generation, and action' as its core.
This approach overcomes the limitations of past 'modality stitching' in multimodal models, achieving an organic fusion of comprehension, generation, and action capabilities at the foundational level.
It also pioneers support for native 8K resolution output, enabling the generation of ultra-high-definition images with color schemes that closely adhere to the given theme.
Most significantly, the model is directly oriented towards 'delivery-grade/production-ready' design.
In challenging professional scenarios such as urban planning, film storyboarding, academic posters, and commercial posters, it aims to propel AI from an auxiliary tool into a new phase of highly usable, agent-driven creation.
Comments