titleSenseNova-U1.5: Towards Native Unified Visual Intelligence
authorsHaiwen Diao, Jiahao Wang, Chenjing Ding, Hanming Deng, Jiangnan Chen, Ruixi Zhang, Ruohui Wang, Wenwen Tong, Xiangyu Fan, Yubo Wang, Yue Zhu, Yuwei Niu, Zhengqi Bai, Zhiqian Lin, Zhitao Yang, Zhongang Cai, Bo Yang, Chen Feng, Chengguang Lv, Guangjia Liu, Guanlin Wang, Hanyu Zhang, Haojia Yu, Hongcan Xiao, Hongli Wang, Huan Wu, Huaping Zhong, Jian Fang, Jianan Fan, Jiaqi Li, Jiefan Lu, Jing Zuo, Jingcheng Ni, Junxiang Xu, Linjun Dai, Mutian Xu, Peishen Yan, Penghao Wu, Ruijie Mao, Ruisi Wang, Shihao Bai, Shuang Yang, Shuya Yang, Shuyan Zheng, Silei Wu, Siying Li, Tao Chu, Tianbo Zhong, Tongxi Zhou, Weichao Luo, Weichen Fan, Wenhao Jia, Wenjie Gao, Xiangli Kong, Yan Li, Yang Yong, Zimo Wen, Zixuan Qian, Wenxiu Sun, Ruihao Gong, Quan Wang, Lewei Lu, Lei Yang, Ziwei Liu, Dahua Lin
date2026-09-10
venuearXiv (preprint, not peer reviewed)
idarXiv:2609.11929
affiliationSenseNova
upvotes145 (Hugging Face Daily Papers, 2026-11-01)
linksabs pdf hf
abstractWe launch SenseNova-U1.5, an 8B-MoT native unified multimodal model that understands, reasons about, and generates visual content within an encoder-free and VAE-free architecture. We strengthen its visual interface through spatially coherent patch reconstruction and scale its training with carefully curated generation and editing data, improved task formulation, structural prompt enhancement, and native resolutions of up to 4K. For post-training, we optimize specialized experts for visual aesthetics, bilingual text rendering, infographic generation, and image editing, and consolidate their capabilities through multi-expert on-policy distillation. Across extensive evaluations, SenseNova-U1.5 largely advances image fidelity, text rendering, complex composition, multi-reference editing, and interleaved generation, while improving instruction following and preserving subject identity, geometry, and unmodified regions. Despite limited exposure to structured formats in its generation data, SenseNova-U1.5 generalizes effectively to long, complex, and structured visual instructions, further proving that multimodal understanding can transfer to visual planning and creation. Together, these findings position native unified modelling as a promising path towards systems that perceive, reason and create within a fully end-to-end framework. We will open-source training code, including supervised fine-tuning, reinforcement learning, and on-policy distillation.

cite

@misc{diao2026sensenova,
  title={SenseNova-U1.5: Towards Native Unified Visual Intelligence},
  author={Haiwen Diao and Jiahao Wang and Chenjing Ding and Hanming Deng and Jiangnan Chen and Ruixi Zhang and Ruohui Wang and Wenwen Tong and Xiangyu Fan and Yubo Wang and Yue Zhu and Yuwei Niu and Zhengqi Bai and Zhiqian Lin and Zhitao Yang and Zhongang Cai and Bo Yang and Chen Feng and Chengguang Lv and Guangjia Liu and Guanlin Wang and Hanyu Zhang and Haojia Yu and Hongcan Xiao and Hongli Wang and Huan Wu and Huaping Zhong and Jian Fang and Jianan Fan and Jiaqi Li and Jiefan Lu and Jing Zuo and Jingcheng Ni and Junxiang Xu and Linjun Dai and Mutian Xu and Peishen Yan and Penghao Wu and Ruijie Mao and Ruisi Wang and Shihao Bai and Shuang Yang and Shuya Yang and Shuyan Zheng and Silei Wu and Siying Li and Tao Chu and Tianbo Zhong and Tongxi Zhou and Weichao Luo and Weichen Fan and Wenhao Jia and Wenjie Gao and Xiangli Kong and Yan Li and Yang Yong and Zimo Wen and Zixuan Qian and Wenxiu Sun and Ruihao Gong and Quan Wang and Lewei Lu and Lei Yang and Ziwei Liu and Dahua Lin},
  year={2026},
  eprint={2609.11929},
  archivePrefix={arXiv},
  url={https://arxiv.org/abs/2609.11929}
}