This new image model is gaining traction: native 4K output, made in China, and open source!

SenseTime has open-sourced SenseNova U1.5-Lite-Preview, a lightweight 8B native unified multimodal model for the community preview.

The model combines language, visual understanding, image generation, and editing in one system, with support for 4K output.

It also handles long prompts, multi-image references, and local precise edits for more controllable image creation.

The release lowers the barrier for producing high-quality visuals and revising images within a single workflow.