SenseNova-Vision-7B-MoT Released

liuziwei7 · x · 2026-07-10

SenseNova-Vision-7B-MoT is now available on ModelScope as a unified multimodal generative model designed for computer vision tasks.

According to the post, it outperforms general vision models across tasks like object detection, semantic segmentation, visual grounding, and depth estimation. It outputs results via text, images, or a mix of both, and was trained on SenseNova-Vision-Corpus-50M under the CC BY-NC 4.0 license.

Related event: SenseNova-Vision Unifies Vision Tasks With 7B-MoT(4 posts)→

Original post →

More from Multimodal

Multimodal channel →