method · first seen 4 Oct 2026
chain-of-thought
Also: CoT, thinking, reasoning, visible CoT, CoT prompting, reasoning trace, thinking tokens
Prompting or training models to produce intermediate step-by-step reasoning before giving a final answer.
Chain-of-thought means having a language model write out intermediate reasoning steps before its final answer, either by prompting or by training. The sources do not introduce the idea. They use it as a base for other work: restarting reasoning chains at chosen steps, compressing the KV cache during long CoT decoding, fixing trace regions in continuous language flows, auditing flawed CoT in vulnerability analysis, and rewarding Swahili reasoning traces with RL. A paper on image generation argues that think blocks act as conditioning rather than plans, and two Hacker News posts discuss reasoning more generally. Nearly all mentions are arXiv papers spread across unrelated groups.
Chain-of-thought is now the default substrate for reasoning models, so work on its faithfulness, cost and structure affects many systems. The rise here reflects ongoing research volume, not a new idea taking hold.
Analyst note (established), 17 Nov 2026.
Mentions per week
PapersCodeDiscussion
Rank 119 on 14 Nov 2026 · 74 mentions in 28 days · 74 independent groups · papers → discussion · first seen 6 weeks ago · acceleration -0.07
Earliest sources
- Expanding LLM ReasoningPapers · 4 Oct 2026 · Rian Atri, Evan Luo
- Spend Bytes on Breadth: Precision-Count Trade-offs for Decode-Time KV Compression in Long Chain-of-Thought ReasoningPapers · 5 Oct 2026 · Runguo Li
- Plan Canvas: Fixed Reasoning Regions for Continuous Language FlowsPapers · 5 Oct 2026 · Miaohe Niu, Pengxiang Li, Jingbo Zhu, Tong Xiao
- Correct Verdicts, Flawed Reasoning: Structured Auditing of LLM-based Vulnerability ReasoningPapers · 5 Oct 2026 · Boyue Caroline Hu, Kaivalya Ahir, Ronghao Ni, Limin Jia
- Quantifying the Stability of Multi-Step Reasoning via Error AmplificationPapers · 5 Oct 2026 · Dongyue Li, Ziniu Zhang, Minxuan Duan, Hongyang R. Zhang
- Base Models Can Reason By Taking a Cue From Training DataPapers · 5 Oct 2026 · Sophie L. Wang, Amil Dravid, Rulin Shao, Kevin Farhat, Sewon Min, Alexei A. Efro…
Latest sources
- Optimal Autoregressive Chain-of-Thought Learning via BaggingPapers · 16 Nov 2026 · Awni Altabaa, John Lafferty, Omar Montasser
- Deep Learning for Physicists: From Linear Regression to Reasoning ModelsPapers · 16 Nov 2026 · Masanori Hanada
- EndoComparer: Comparative Surgical Video Question Answering with Pseudo-Siamese Visual Adaptation and Perceive-then-Compare ReasoningPapers · 14 Nov 2026 · Shi Li, Lalith Sharan, Joël L. Lavanchy, Kun Yuan, Lorenzo Arboit, Nabani Banik…
- A Think Block Is Conditioning, Not a PlanPapers · 14 Nov 2026 · Jian Xu, Delu Zeng, Qibin Zhao
- Show Your Work in Swahili: Reinforcement Learning for Native-Language Mathematical ReasoningPapers · 11 Nov 2026 · Jessica Lundin, Katherine Rosenfeld, David Adelani, Geoffrey Mosoti Nyakiongora,…
- Same Task, Different Statement: When and How Reformulation Helps LLM Problem SolvingPapers · 11 Nov 2026 · Wentao Zhang, Xiaoyu Zhang, Chengwei Liu, Yutao Hu, Tao Zhang, Junsheng Wu, Hong…
- What objective is reasoning solving?Discussion · 10 Nov 2026 · gmays · 162 points
- Code as Explanations: Executable and Verifiable LLM ExplanationsPapers · 10 Nov 2026 · Nhi Nguyen, Shauli Ravfogel, Rajesh Ranganath
- Cognitive Mapping of Learning by Thinking Mechanisms in Large Language ModelsPapers · 10 Nov 2026 · Dong Han, Jin Zhang, Kunyue Yang
- Breakdown of Autoregressive Multi-Step Reasoning Under Concentrated vs. Diffuse Attention Head DropoutPapers · 10 Nov 2026 · Karol Struniawski
- How Post-Training Influences Verbalized Algorithmic Computation in Language ModelsPapers · 10 Nov 2026 · Niall Dalton
- Reasoning-Aware Relational Representation Learning for Open-Vocabulary Scene Graph GenerationPapers · 10 Nov 2026 · Jiawei Xiao, Boya Wang, Longtian Qiu, Shan Ning, Xuming He