_version_ 1866915381448278016
author Zhu, Rui-Jie
Peng, Tianhao
Cheng, Tianhao
Qu, Xingwei
Huang, Jinfa
Zhu, Dawei
Wang, Hao
Xue, Kaiwen
Zhang, Xuanliang
Shan, Yong
Cai, Tianle
Kergan, Taylor
Kembay, Assel
Smith, Andrew
Lin, Chenghua
Nguyen, Binh
Pan, Yuqi
Chou, Yuhong
Cai, Zefan
Wu, Zhenhe
Zhao, Yongchi
Liu, Tianyu
Yang, Jian
Zhou, Wangchunshu
Zheng, Chujie
Li, Chongxuan
Zhou, Yuyin
Li, Zhoujun
Zhang, Zhaoxiang
Liu, Jiaheng
Zhang, Ge
Huang, Wenhao
Eshraghian, Jason
author_facet Zhu, Rui-Jie
Peng, Tianhao
Cheng, Tianhao
Qu, Xingwei
Huang, Jinfa
Zhu, Dawei
Wang, Hao
Xue, Kaiwen
Zhang, Xuanliang
Shan, Yong
Cai, Tianle
Kergan, Taylor
Kembay, Assel
Smith, Andrew
Lin, Chenghua
Nguyen, Binh
Pan, Yuqi
Chou, Yuhong
Cai, Zefan
Wu, Zhenhe
Zhao, Yongchi
Liu, Tianyu
Yang, Jian
Zhou, Wangchunshu
Zheng, Chujie
Li, Chongxuan
Zhou, Yuyin
Li, Zhoujun
Zhang, Zhaoxiang
Liu, Jiaheng
Zhang, Ge
Huang, Wenhao
Eshraghian, Jason
contents Large Language Models (LLMs) have demonstrated impressive reasoning capabilities, especially when guided by explicit chain-of-thought (CoT) reasoning that verbalizes intermediate steps. While CoT improves both interpretability and accuracy, its dependence on natural language reasoning limits the model's expressive bandwidth. Latent reasoning tackles this bottleneck by performing multi-step inference entirely in the model's continuous hidden state, eliminating token-level supervision. To advance latent reasoning research, this survey provides a comprehensive overview of the emerging field of latent reasoning. We begin by examining the foundational role of neural network layers as the computational substrate for reasoning, highlighting how hierarchical representations support complex transformations. Next, we explore diverse latent reasoning methodologies, including activation-based recurrence, hidden state propagation, and fine-tuning strategies that compress or internalize explicit reasoning traces. Finally, we discuss advanced paradigms such as infinite-depth latent reasoning via masked diffusion models, which enable globally consistent and reversible reasoning processes. By unifying these perspectives, we aim to clarify the conceptual landscape of latent reasoning and chart future directions for research at the frontier of LLM cognition. An associated GitHub repository collecting the latest papers and repos is available at: https://github.com/multimodal-art-projection/LatentCoT-Horizon/.
format Preprint
id arxiv_https___arxiv_org_abs_2507_06203
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Survey on Latent Reasoning
Zhu, Rui-Jie
Peng, Tianhao
Cheng, Tianhao
Qu, Xingwei
Huang, Jinfa
Zhu, Dawei
Wang, Hao
Xue, Kaiwen
Zhang, Xuanliang
Shan, Yong
Cai, Tianle
Kergan, Taylor
Kembay, Assel
Smith, Andrew
Lin, Chenghua
Nguyen, Binh
Pan, Yuqi
Chou, Yuhong
Cai, Zefan
Wu, Zhenhe
Zhao, Yongchi
Liu, Tianyu
Yang, Jian
Zhou, Wangchunshu
Zheng, Chujie
Li, Chongxuan
Zhou, Yuyin
Li, Zhoujun
Zhang, Zhaoxiang
Liu, Jiaheng
Zhang, Ge
Huang, Wenhao
Eshraghian, Jason
Computation and Language
Large Language Models (LLMs) have demonstrated impressive reasoning capabilities, especially when guided by explicit chain-of-thought (CoT) reasoning that verbalizes intermediate steps. While CoT improves both interpretability and accuracy, its dependence on natural language reasoning limits the model's expressive bandwidth. Latent reasoning tackles this bottleneck by performing multi-step inference entirely in the model's continuous hidden state, eliminating token-level supervision. To advance latent reasoning research, this survey provides a comprehensive overview of the emerging field of latent reasoning. We begin by examining the foundational role of neural network layers as the computational substrate for reasoning, highlighting how hierarchical representations support complex transformations. Next, we explore diverse latent reasoning methodologies, including activation-based recurrence, hidden state propagation, and fine-tuning strategies that compress or internalize explicit reasoning traces. Finally, we discuss advanced paradigms such as infinite-depth latent reasoning via masked diffusion models, which enable globally consistent and reversible reasoning processes. By unifying these perspectives, we aim to clarify the conceptual landscape of latent reasoning and chart future directions for research at the frontier of LLM cognition. An associated GitHub repository collecting the latest papers and repos is available at: https://github.com/multimodal-art-projection/LatentCoT-Horizon/.
title A Survey on Latent Reasoning
topic Computation and Language
url https://arxiv.org/abs/2507.06203