_version_ 1866910008125423616
author Cao, Siyu
Chen, Hangting
Chen, Peng
Cheng, Yiji
Cui, Yutao
Deng, Xinchi
Dong, Ying
Gong, Kipper
Gu, Tianpeng
Gu, Xiusen
Hang, Tiankai
Huang, Duojun
Jiang, Jie
Jiang, Zhengkai
Kong, Weijie
Li, Changlin
Li, Donghao
Li, Junzhe
Li, Xin
Li, Yang
Li, Zhenxi
Li, Zhimin
Lin, Jiaxin
Linus
Liu, Lucaz
Liu, Shu
Liu, Songtao
Liu, Yu
Liu, Yuhong
Long, Yanxin
Lu, Fanbin
Lu, Qinglin
Peng, Yuyang
Peng, Yuanbo
Shen, Xiangwei
Shi, Yixuan
Tao, Jiale
Tao, Yangyu
Tian, Qi
Wan, Pengfei
Wang, Chunyu
Wang, Kai
Wang, Lei
Wang, Linqing
Wang, Lucas
Wang, Qixun
Wang, Weiyan
Wen, Hao
Wu, Bing
Wu, Jianbing
Wu, Yue
Xie, Senhao
Yang, Fang
Yang, Miles
Yang, Xiaofeng
Yang, Xuan
Yang, Zhantao
Yu, Jingmiao
Yuan, Zheng
Zhang, Chao
Zhang, Jian-Wei
Zhang, Peizhen
Zhang, Shi-Xue
Zhang, Tao
Zhang, Weigang
Zhang, Yepeng
Zhang, Yingfang
Zhang, Zihao
Zhang, Zijian
Zhao, Penghao
Zhao, Zhiyuan
Zhe, Xuefei
Zhu, Jianchen
Zhong, Zhao
author_facet Cao, Siyu
Chen, Hangting
Chen, Peng
Cheng, Yiji
Cui, Yutao
Deng, Xinchi
Dong, Ying
Gong, Kipper
Gu, Tianpeng
Gu, Xiusen
Hang, Tiankai
Huang, Duojun
Jiang, Jie
Jiang, Zhengkai
Kong, Weijie
Li, Changlin
Li, Donghao
Li, Junzhe
Li, Xin
Li, Yang
Li, Zhenxi
Li, Zhimin
Lin, Jiaxin
Linus
Liu, Lucaz
Liu, Shu
Liu, Songtao
Liu, Yu
Liu, Yuhong
Long, Yanxin
Lu, Fanbin
Lu, Qinglin
Peng, Yuyang
Peng, Yuanbo
Shen, Xiangwei
Shi, Yixuan
Tao, Jiale
Tao, Yangyu
Tian, Qi
Wan, Pengfei
Wang, Chunyu
Wang, Kai
Wang, Lei
Wang, Linqing
Wang, Lucas
Wang, Qixun
Wang, Weiyan
Wen, Hao
Wu, Bing
Wu, Jianbing
Wu, Yue
Xie, Senhao
Yang, Fang
Yang, Miles
Yang, Xiaofeng
Yang, Xuan
Yang, Zhantao
Yu, Jingmiao
Yuan, Zheng
Zhang, Chao
Zhang, Jian-Wei
Zhang, Peizhen
Zhang, Shi-Xue
Zhang, Tao
Zhang, Weigang
Zhang, Yepeng
Zhang, Yingfang
Zhang, Zihao
Zhang, Zijian
Zhao, Penghao
Zhao, Zhiyuan
Zhe, Xuefei
Zhu, Jianchen
Zhong, Zhao
contents We present HunyuanImage 3.0, a native multimodal model that unifies multimodal understanding and generation within an autoregressive framework, with its image generation module publicly available. The achievement of HunyuanImage 3.0 relies on several key components, including meticulous data curation, advanced architecture design, a native Chain-of-Thoughts schema, progressive model pre-training, aggressive model post-training, and an efficient infrastructure that enables large-scale training and inference. With these advancements, we successfully trained a Mixture-of-Experts (MoE) model comprising over 80 billion parameters in total, with 13 billion parameters activated per token during inference, making it the largest and most powerful open-source image generative model to date. We conducted extensive experiments and the results of automatic and human evaluation of text-image alignment and visual quality demonstrate that HunyuanImage 3.0 rivals previous state-of-the-art models. By releasing the code and weights of HunyuanImage 3.0, we aim to enable the community to explore new ideas with a state-of-the-art foundation model, fostering a dynamic and vibrant multimodal ecosystem. All open source assets are publicly available at https://github.com/Tencent-Hunyuan/HunyuanImage-3.0
format Preprint
id arxiv_https___arxiv_org_abs_2509_23951
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle HunyuanImage 3.0 Technical Report
Cao, Siyu
Chen, Hangting
Chen, Peng
Cheng, Yiji
Cui, Yutao
Deng, Xinchi
Dong, Ying
Gong, Kipper
Gu, Tianpeng
Gu, Xiusen
Hang, Tiankai
Huang, Duojun
Jiang, Jie
Jiang, Zhengkai
Kong, Weijie
Li, Changlin
Li, Donghao
Li, Junzhe
Li, Xin
Li, Yang
Li, Zhenxi
Li, Zhimin
Lin, Jiaxin
Linus
Liu, Lucaz
Liu, Shu
Liu, Songtao
Liu, Yu
Liu, Yuhong
Long, Yanxin
Lu, Fanbin
Lu, Qinglin
Peng, Yuyang
Peng, Yuanbo
Shen, Xiangwei
Shi, Yixuan
Tao, Jiale
Tao, Yangyu
Tian, Qi
Wan, Pengfei
Wang, Chunyu
Wang, Kai
Wang, Lei
Wang, Linqing
Wang, Lucas
Wang, Qixun
Wang, Weiyan
Wen, Hao
Wu, Bing
Wu, Jianbing
Wu, Yue
Xie, Senhao
Yang, Fang
Yang, Miles
Yang, Xiaofeng
Yang, Xuan
Yang, Zhantao
Yu, Jingmiao
Yuan, Zheng
Zhang, Chao
Zhang, Jian-Wei
Zhang, Peizhen
Zhang, Shi-Xue
Zhang, Tao
Zhang, Weigang
Zhang, Yepeng
Zhang, Yingfang
Zhang, Zihao
Zhang, Zijian
Zhao, Penghao
Zhao, Zhiyuan
Zhe, Xuefei
Zhu, Jianchen
Zhong, Zhao
Computer Vision and Pattern Recognition
We present HunyuanImage 3.0, a native multimodal model that unifies multimodal understanding and generation within an autoregressive framework, with its image generation module publicly available. The achievement of HunyuanImage 3.0 relies on several key components, including meticulous data curation, advanced architecture design, a native Chain-of-Thoughts schema, progressive model pre-training, aggressive model post-training, and an efficient infrastructure that enables large-scale training and inference. With these advancements, we successfully trained a Mixture-of-Experts (MoE) model comprising over 80 billion parameters in total, with 13 billion parameters activated per token during inference, making it the largest and most powerful open-source image generative model to date. We conducted extensive experiments and the results of automatic and human evaluation of text-image alignment and visual quality demonstrate that HunyuanImage 3.0 rivals previous state-of-the-art models. By releasing the code and weights of HunyuanImage 3.0, we aim to enable the community to explore new ideas with a state-of-the-art foundation model, fostering a dynamic and vibrant multimodal ecosystem. All open source assets are publicly available at https://github.com/Tencent-Hunyuan/HunyuanImage-3.0
title HunyuanImage 3.0 Technical Report
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2509.23951