推理模型论文趋势 2023–2026

主要结论

• 推理模型论文从 2023 年的 470 篇增长到 2026 年的 1,799 篇，三年间扩大到约 3.8 倍。
• NeurIPS 是发推理模型论文最多的会议，2023–2026 年共 1,253 篇，占四个会议总量的 36%。
• 交叉方向里涨得最快的是强化学习：同时属于两个主题的论文从 2023 年的 9 篇增至 2026 年的 437 篇。
• 在「前沿系统」的 4 个主题里，它的 2026 年论文数排第 3；同组增长最快的AI 智能体是 12 倍，推理模型是 3.8 倍。

代表论文

按年份和会议均匀挑选，方便快速看到每一届的样貌

• NeurIPS 2026A Mechanistic Analysis of Looped Reasoning Language Models
• ICML 2026A Formal Comparison Between Chain of Thought and Latent Thought
• ICLR 2026A Simple "Motivation" Can Enhance Reinforcement Finetuning of Large Reasoning Models
• CVPR 2026Asking like Socrates Socrates helps VLMs understand remote sensing images
• NeurIPS 2025A Frustratingly Simple Yet Highly Effective Attack Baseline: Over 90% Success Rate Against the Strong Black-box Models of GPT-4.5/4o/o1
• ICML 2025A Reasoning-Based Approach to Cryptic Crossword Clue Solving
• ICLR 2025A Formal Framework for Understanding Length Generalization in Transformers
• CVPR 2025Argus: Vision-Centric Reasoning with Grounded Chain-of-Thought
• NeurIPS 2024Are Large Language Models Good Statisticians?
• ICML 2024Algorithm of Thoughts: Enhancing Exploration of Ideas in Large Language Models
• ICLR 2024Amortizing intractable inference in large language models
• CVPR 2024Generate Subgoal Images before Act: Unlocking the Chain-of-Thought Reasoning in Diffusion Model for Robot Manipulation with Multimodal Prompts
• NeurIPS 2023A benchmark of categorical encoders for binary classification
• ICML 2023Can We Scale Transformers to Predict Parameters of Diverse ImageNet Models?
• ICLR 2023Automatic Chain of Thought Prompting in Large Language Models
• CVPR 2023ActMAD: Activation Matching To Align Distributions for Test-Time-Training
• NeurIPS 2026A Solvable Model of Chain-of-Thought in In-Context Learning
• ICML 2026A Narrowing Geometry in Contaminated Reasoning
• ICLR 2026A State-Transition Framework for Efficient LLM Reasoning
• CVPR 2026Beyond Weak Supervision MLLMs-Guided Graded Knowledge Distillation for Unsupervised Camouflaged

全部推理模型论文

共 203 篇，第 3 / 5 页

101. ICLR 2025Matcha: Mitigating Graph Structure Shifts with Test-Time Adaptation
102. ICLR 2025MathCoder2: Better Math Reasoning from Continued Pretraining on Model-translated Mathematical Code
103. ICLR 2025MathGAP: Out-of-Distribution Evaluation on Problems with Arbitrarily Complex Proofs
104. ICLR 2025MAVIS: Mathematical Visual Instruction Tuning with an Automatic Data Engine
105. ICLR 2025Mechanism and Emergence of Stacked Attention Heads in Multi-Layer Transformers
106. ICLR 2025metabench - A Sparse Benchmark of Reasoning and Knowledge in Large Language Models
107. ICLR 2025Mind the Gap: Examining the Self-Improvement Capabilities of Large Language Models
108. ICLR 2025Mind the GAP: Glimpse-based Active Perception improves generalization and sample efficiency of visual reasoning
109. ICLR 2025MIND: Math Informed syNthetic Dialogues for Pretraining LLMs
110. ICLR 2025MIRAGE: Evaluating and Explaining Inductive Reasoning Process in Language Models
111. ICLR 2025Mitigating Modality Prior-Induced Hallucinations in Multimodal Large Language Models via Deciphering Attention Causality
112. ICLR 2025Mixture of Parrots: Experts improve memorization more than reasoning
113. ICLR 2025MLPs Learn In-Context on Regression and Classification Tasks
114. ICLR 2025MMAU: A Massive Multi-Task Audio Understanding and Reasoning Benchmark
115. ICLR 2025MMR: A Large-scale Benchmark Dataset for Multi-target and Multi-granularity Reasoning Segmentation
116. ICLR 2025Mutual Reasoning Makes Smaller LLMs Stronger Problem-Solver
117. ICLR 2025NL-Eye: Abductive NLI For Images
118. ICLR 2025Node-Time Conditional Prompt Learning in Dynamic Graphs
119. ICLR 2025Non-myopic Generation of Language Models for Reasoning and Planning
120. ICLR 2025Not All Heads Matter: A Head-Level KV Cache Compression Method with Integrated Retrieval and Reasoning
121. ICLR 2025Number Cookbook: Number Understanding of Language Models and How to Improve It
122. ICLR 2025NutriBench: A Dataset for Evaluating Large Language Models in Nutrition Estimation from Meal Descriptions
123. ICLR 2025OCEAN: Offline Chain-of-thought Evaluation and Alignment in Large Language Models
124. ICLR 2025OmnixR: Evaluating Omni-modality Language Models on Reasoning across Modalities
125. ICLR 2025On the Adversarial Risk of Test Time Adaptation: An Investigation into Realistic Test-Time Data Poisoning
126. ICLR 2025On the Adversarial Vulnerability of Label-Free Test-Time Adaptation
127. ICLR 2025On the self-verification limitations of large language models on reasoning and planning tasks
128. ICLR 2025OSCAR: Operating System Control via State-Aware Reasoning and Re-Planning
129. ICLR 2025Personality Alignment of Large Language Models
130. ICLR 2025Physics of Language Models: Part 2.1, Grade-School Math and the Hidden Reasoning Process
131. ICLR 2025Planning in Natural Language Improves LLM Search for Code Generation
132. ICLR 2025Preference Optimization for Reasoning with Pseudo Feedback
133. ICLR 2025Preserving Diversity in Supervised Fine-Tuning of Large Language Models
134. ICLR 2025Probing the Latent Hierarchical Structure of Data via Diffusion Models
135. ICLR 2025Procedural Knowledge in Pretraining Drives Reasoning in Large Language Models
136. ICLR 2025Process Reward Model with Q-value Rankings
137. ICLR 2025Reasoning Elicitation in Language Models via Counterfactual Feedback
138. ICLR 2025Reasoning of Large Language Models over Knowledge Graphs with Super-Relations
139. ICLR 2025Reasoning with Latent Thoughts: On the Power of Looped Transformers
140. ICLR 2025RefactorBench: Evaluating Stateful Reasoning in Language Agents Through Code
141. ICLR 2025ReGenesis: LLMs can Grow into Reasoning Generalists via Self-Improvement
142. ICLR 2025Retrieval Head Mechanistically Explains Long-Context Factuality
143. ICLR 2025Revisiting Random Walks for Learning on Graphs
144. ICLR 2025Rewarding Progress: Scaling Automated Process Verifiers for LLM Reasoning
145. ICLR 2025RNNs are not Transformers (Yet): The Key Bottleneck on In-Context Retrieval
146. ICLR 2025RocketEval: Efficient automated LLM evaluation via grading checklist
147. ICLR 2025SBSC: Step-by-Step Coding for Improving Mathematical Olympiad Performance
148. ICLR 2025Scaling Large Language Model-based Multi-Agent Collaboration
149. ICLR 2025Scaling LLM Test-Time Compute Optimally Can be More Effective than Scaling Parameters for Reasoning
150. ICLR 2025Searching for Optimal Solutions with LLMs via Bayesian Optimization

常见问题

2023–2026 年四大会议一共有多少篇推理模型论文？

共 3,512 篇，按年份分别是 2023 年 470 篇、2024 年 311 篇、2025 年 932 篇、2026 年 1,799 篇。

NeurIPS 2026 有多少篇推理模型论文？

NeurIPS 2026 有 361 篇推理模型论文，上一届是 500 篇。

ICLR 2026 有多少篇推理模型论文？

ICLR 2026 有 652 篇推理模型论文，上一届是 203 篇。

ICML 2026 有多少篇推理模型论文？

ICML 2026 有 582 篇推理模型论文，上一届是 159 篇。

CVPR 2026 有多少篇推理模型论文？

CVPR 2026 有 204 篇推理模型论文，上一届是 70 篇。

哪个会议的推理模型论文最多？

NeurIPS，2023–2026 年共 1,253 篇。