SenseTime Open-Sources Lightweight Multimodal Model SenseNova U1.5-Lite-Preview

SenseTime recently announced the open-sourcing of the preview version of its lightweight unified multimodal model, SenseNova U1.5-Lite-Preview. This model significantly outperforms its predecessor U1 across multiple mainstream generation and editing benchmarks, achieving image generation and editing capabilities comparable to commercial closed-source models at a lightweight scale of only 8B-MoT.

Compared to U1, U1.5-Lite-Preview natively supports 4K image generation, offers finer local textures, details, and realistic world textures, improves Chinese and English text generation and complex layout organization, as well as more stable image editing and visual instruction following.

The model weights are now available on GitHub, Hugging Face, and ModelScope, allowing developers to download and use them freely. This open-source release is expected to promote the application of multimodal models on lightweight devices and lower the barrier for developers.