SenseTime today announced the preview version of SenseNova U1.5-Lite-Preview, a lightweight unified multimodal model open-sourced for the community.
According to the announcement, thanks to enhancements in multiple capabilities, U1.5-Lite-Preview significantly outperforms U1 on several mainstream generation and editing quality benchmarks. With its lightweight 8B-MoT scale, it delivers image generation capabilities and editing results comparable to those of commercial closed-source models.
The company stated that, compared with U1, U1.5-Lite-Preview delivers significant improvements in the following areas:
Native support for 4K image generation;
More refined local textures, details, and real-world visual fidelity;
More accurate Chinese and English text generation and complex layout organization;
More stable image editing and visual instruction following.
Open-source links:
GitHub: https://github.com/OpenSenseNova/SenseNova-U1/blob/main/docs/u1.5_preview.md
Hugging Face: https://huggingface.co/sensenova/SenseNova-U1.5-8B-MoT-Preview
ModelScope: https://modelscope.cn/models/SenseNova/SenseNova-U1.5-8B-MoT-Preview