Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | https://arxiv.org/abs/2507.01006 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911349276147712 |
|---|---|
| author | V Team Hong, Wenyi Yu, Wenmeng Gu, Xiaotao Wang, Guo Gan, Guobing Tang, Haomiao Cheng, Jiale Qi, Ji Ji, Junhui Pan, Lihang Duan, Shuaiqi Wang, Weihan Wang, Yan Cheng, Yean He, Zehai Su, Zhe Yang, Zhen Pan, Ziyang Zeng, Aohan Wang, Baoxu Chen, Bin Shi, Boyan Pang, Changyu Zhang, Chenhui Yin, Da Yang, Fan Chen, Guoqing Li, Haochen Zhu, Jiale Chen, Jiali Xu, Jiaxing Xu, Jiazheng Chen, Jing Lin, Jinghao Chen, Jinhao Wang, Jinjiang Chen, Junjie Lei, Leqi Gong, Letian Pan, Leyi Liu, Mingdao Xu, Mingde Zhang, Mingzhi Zheng, Qinkai Lyu, Ruiliang Tu, Shangqin Yang, Sheng Meng, Shengbiao Zhong, Shi Huang, Shiyu Zhao, Shuyuan Xue, Siyan Zhang, Tianshu Luo, Tianwei Hao, Tianxiang Tong, Tianyu Jia, Wei Li, Wenkai Liu, Xiao Zhang, Xiaohan Lyu, Xin Zhang, Xinyu Fan, Xinyue Huang, Xuancheng Xue, Yadong Wang, Yanfeng Wang, Yanling Wang, Yanzi An, Yifan Du, Yifan Huang, Yiheng Niu, Yilin Shi, Yiming Wang, Yu Wang, Yuan Yue, Yuanchang Li, Yuchen Liu, Yusen Zhang, Yutao Wang, Yuting Zhang, Yuxuan Xue, Zhao Du, Zhengxiao Hou, Zhenyu Wang, Zihan Zhang, Peng Liu, Debing Xu, Bin Li, Juanzi Huang, Minlie Dong, Yuxiao Tang, Jie |
| author_facet | V Team Hong, Wenyi Yu, Wenmeng Gu, Xiaotao Wang, Guo Gan, Guobing Tang, Haomiao Cheng, Jiale Qi, Ji Ji, Junhui Pan, Lihang Duan, Shuaiqi Wang, Weihan Wang, Yan Cheng, Yean He, Zehai Su, Zhe Yang, Zhen Pan, Ziyang Zeng, Aohan Wang, Baoxu Chen, Bin Shi, Boyan Pang, Changyu Zhang, Chenhui Yin, Da Yang, Fan Chen, Guoqing Li, Haochen Zhu, Jiale Chen, Jiali Xu, Jiaxing Xu, Jiazheng Chen, Jing Lin, Jinghao Chen, Jinhao Wang, Jinjiang Chen, Junjie Lei, Leqi Gong, Letian Pan, Leyi Liu, Mingdao Xu, Mingde Zhang, Mingzhi Zheng, Qinkai Lyu, Ruiliang Tu, Shangqin Yang, Sheng Meng, Shengbiao Zhong, Shi Huang, Shiyu Zhao, Shuyuan Xue, Siyan Zhang, Tianshu Luo, Tianwei Hao, Tianxiang Tong, Tianyu Jia, Wei Li, Wenkai Liu, Xiao Zhang, Xiaohan Lyu, Xin Zhang, Xinyu Fan, Xinyue Huang, Xuancheng Xue, Yadong Wang, Yanfeng Wang, Yanling Wang, Yanzi An, Yifan Du, Yifan Huang, Yiheng Niu, Yilin Shi, Yiming Wang, Yu Wang, Yuan Yue, Yuanchang Li, Yuchen Liu, Yusen Zhang, Yutao Wang, Yuting Zhang, Yuxuan Xue, Zhao Du, Zhengxiao Hou, Zhenyu Wang, Zihan Zhang, Peng Liu, Debing Xu, Bin Li, Juanzi Huang, Minlie Dong, Yuxiao Tang, Jie |
| contents | We present GLM-4.1V-Thinking, GLM-4.5V, and GLM-4.6V, a family of vision-language models (VLMs) designed to advance general-purpose multimodal understanding and reasoning. In this report, we share our key findings in the development of the reasoning-centric training framework. We first develop a capable vision foundation model with significant potential through large-scale pre-training, which arguably sets the upper bound for the final performance. We then propose Reinforcement Learning with Curriculum Sampling (RLCS) to unlock the full potential of the model, leading to comprehensive capability enhancement across a diverse range of tasks, including STEM problem solving, video understanding, content recognition, coding, grounding, GUI-based agents, and long document interpretation. In a comprehensive evaluation across 42 public benchmarks, GLM-4.5V achieves state-of-the-art performance on nearly all tasks among open-source models of similar size, and demonstrates competitive or even superior results compared to closed-source models such as Gemini-2.5-Flash on challenging tasks including Coding and GUI Agents. Meanwhile, the smaller GLM-4.1V-9B-Thinking remains highly competitive-achieving superior results to the much larger Qwen2.5-VL-72B on 29 benchmarks. We open-source both GLM-4.1V-9B-Thinking and GLM-4.5V. We further introduce the GLM-4.6V series, open-source multimodal models with native tool use and a 128K context window. A brief overview is available at https://z.ai/blog/glm-4.6v. Code, models and more information are released at https://github.com/zai-org/GLM-V. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_01006 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | GLM-4.5V and GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning V Team Hong, Wenyi Yu, Wenmeng Gu, Xiaotao Wang, Guo Gan, Guobing Tang, Haomiao Cheng, Jiale Qi, Ji Ji, Junhui Pan, Lihang Duan, Shuaiqi Wang, Weihan Wang, Yan Cheng, Yean He, Zehai Su, Zhe Yang, Zhen Pan, Ziyang Zeng, Aohan Wang, Baoxu Chen, Bin Shi, Boyan Pang, Changyu Zhang, Chenhui Yin, Da Yang, Fan Chen, Guoqing Li, Haochen Zhu, Jiale Chen, Jiali Xu, Jiaxing Xu, Jiazheng Chen, Jing Lin, Jinghao Chen, Jinhao Wang, Jinjiang Chen, Junjie Lei, Leqi Gong, Letian Pan, Leyi Liu, Mingdao Xu, Mingde Zhang, Mingzhi Zheng, Qinkai Lyu, Ruiliang Tu, Shangqin Yang, Sheng Meng, Shengbiao Zhong, Shi Huang, Shiyu Zhao, Shuyuan Xue, Siyan Zhang, Tianshu Luo, Tianwei Hao, Tianxiang Tong, Tianyu Jia, Wei Li, Wenkai Liu, Xiao Zhang, Xiaohan Lyu, Xin Zhang, Xinyu Fan, Xinyue Huang, Xuancheng Xue, Yadong Wang, Yanfeng Wang, Yanling Wang, Yanzi An, Yifan Du, Yifan Huang, Yiheng Niu, Yilin Shi, Yiming Wang, Yu Wang, Yuan Yue, Yuanchang Li, Yuchen Liu, Yusen Zhang, Yutao Wang, Yuting Zhang, Yuxuan Xue, Zhao Du, Zhengxiao Hou, Zhenyu Wang, Zihan Zhang, Peng Liu, Debing Xu, Bin Li, Juanzi Huang, Minlie Dong, Yuxiao Tang, Jie Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning We present GLM-4.1V-Thinking, GLM-4.5V, and GLM-4.6V, a family of vision-language models (VLMs) designed to advance general-purpose multimodal understanding and reasoning. In this report, we share our key findings in the development of the reasoning-centric training framework. We first develop a capable vision foundation model with significant potential through large-scale pre-training, which arguably sets the upper bound for the final performance. We then propose Reinforcement Learning with Curriculum Sampling (RLCS) to unlock the full potential of the model, leading to comprehensive capability enhancement across a diverse range of tasks, including STEM problem solving, video understanding, content recognition, coding, grounding, GUI-based agents, and long document interpretation. In a comprehensive evaluation across 42 public benchmarks, GLM-4.5V achieves state-of-the-art performance on nearly all tasks among open-source models of similar size, and demonstrates competitive or even superior results compared to closed-source models such as Gemini-2.5-Flash on challenging tasks including Coding and GUI Agents. Meanwhile, the smaller GLM-4.1V-9B-Thinking remains highly competitive-achieving superior results to the much larger Qwen2.5-VL-72B on 29 benchmarks. We open-source both GLM-4.1V-9B-Thinking and GLM-4.5V. We further introduce the GLM-4.6V series, open-source multimodal models with native tool use and a 128K context window. A brief overview is available at https://z.ai/blog/glm-4.6v. Code, models and more information are released at https://github.com/zai-org/GLM-V. |
| title | GLM-4.5V and GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2507.01006 |