推理模型论文趋势 2023–2026

链式推理、测试时计算与显式推理能力

统计范围：NeurIPS、ICLR、ICML、CVPR 四个会议，2023–2026 年，共 3,512 篇被归入该主题的论文。

主要结论

每年论文数

各会议论文数

按年份和会议统计

推理模型论文数，按年份和会议
年份NeurIPSICLRICMLCVPR合计
20232715710933470
2024121846343311
202550020315970932
20263616525822041,799

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

全部推理模型论文

年份全部2023202420252026
会议全部NeurIPSICLRICMLCVPR

共 1,253 篇，第 17 / 26 页

  1. NeurIPS 2025Thinking vs. Doing: Improving Agent Reasoning by Scaling Test-Time Interaction
  2. NeurIPS 2025Thinkless: LLM Learns When to Think
  3. NeurIPS 2025ThinkSound: Chain-of-Thought Reasoning in Multimodal LLMs for Audio Generation and Editing
  4. NeurIPS 2025Thoughts Are All Over the Place: On the Underthinking of Long Reasoning Models
  5. NeurIPS 2025TimE: A Multi-level Benchmark for Temporal Reasoning of LLMs in Real-World Scenarios
  6. NeurIPS 2025Titans: Learning to Memorize at Test Time
  7. NeurIPS 2025TokenSqueeze: Performance-Preserving Compression for Reasoning LLMs
  8. NeurIPS 2025TOMCAT: Test-time Comprehensive Knowledge Accumulation for Compositional Zero-Shot Learning
  9. NeurIPS 2025Too Late to Recall: Explaining the Two-Hop Problem in Multimodal Knowledge Retrieval
  10. NeurIPS 2025Tool-Augmented Spatiotemporal Reasoning for Streamlining Video Question Answering Task
  11. NeurIPS 2025Topology of Reasoning: Understanding Large Reasoning Models through Reasoning Graph Properties
  12. NeurIPS 2025TopoPoint: Enhance Topology Reasoning via Endpoint Detection in Autonomous Driving
  13. NeurIPS 2025Towards Thinking-Optimal Scaling of Test-Time Compute for LLM Reasoning
  14. NeurIPS 2025Tracking and Understanding Object Transformations
  15. NeurIPS 2025Training a Scientific Reasoning Model for Chemistry
  16. NeurIPS 2025Training Language Models to Reason Efficiently
  17. NeurIPS 2025Trajectory Balance with Asynchrony: Decoupling Exploration and Learning for Fast, Scalable LLM Post-Training
  18. NeurIPS 2025Transformers Provably Learn Chain-of-Thought Reasoning with Length Generalization
  19. NeurIPS 2025Tropical Attention: Neural Algorithmic Reasoning for Combinatorial Algorithms
  20. NeurIPS 2025Trust, But Verify: A Self-Verification Approach to Reinforcement Learning with Verifiable Rewards
  21. NeurIPS 2025TRUST: Test-Time Refinement using Uncertainty-Guided SSM Traverses
  22. NeurIPS 2025Two Experts Are All You Need for Steering Thinking: Reinforcing Cognitive Effort in MoE Reasoning Models Without Additional Training
  23. NeurIPS 2025UFT: Unifying Supervised and Reinforcement Fine-Tuning
  24. NeurIPS 2025Understanding and Mitigating Numerical Sources of Nondeterminism in LLM Inference
  25. NeurIPS 2025Uni-MuMER: Unified Multi-Task Fine-Tuning of Vision-Language Model for Handwritten Mathematical Expression Recognition
  26. NeurIPS 2025Unified Multimodal Chain-of-Thought Reward Model through Reinforcement Fine-Tuning
  27. NeurIPS 2025UniGen: Enhanced Training & Test-Time Strategies for Unified Multimodal Understanding and Generation
  28. NeurIPS 2025UniPixel: Unified Object Referring and Segmentation for Pixel-Level Visual Reasoning
  29. NeurIPS 2025UniTransfer: Video Concept Transfer via Progressive Spatio-Temporal Decomposition
  30. NeurIPS 2025Unlabeled Data Can Provably Enhance In-Context Learning of Transformers
  31. NeurIPS 2025Unlocking Multimodal Mathematical Reasoning via Process Reward Model
  32. NeurIPS 2025Unveiling Chain of Step Reasoning for Vision-Language Models with Fine-grained Rewards
  33. NeurIPS 2025Unveiling the Compositional Ability Gap in Vision-Language Reasoning Model
  34. NeurIPS 2025Vad-R1: Towards Video Anomaly Reasoning via Perception-to-Cognition Chain-of-Thought
  35. NeurIPS 2025Value-Guided Search for Efficient Chain-of-Thought Reasoning
  36. NeurIPS 2025VeriThinker: Learning to Verify Makes Reasoning Model Efficient
  37. NeurIPS 2025VeriThoughts: Enabling Automated Verilog Code Generation using Reasoning and Formal Verification
  38. NeurIPS 2025ViCrit: A Verifiable Reinforcement Learning Proxy Task for Visual Perception in VLMs
  39. NeurIPS 2025Video-R1: Reinforcing Video Reasoning in MLLMs
  40. NeurIPS 2025VideoChat-R1.5: Visual Test-Time Scaling to Reinforce Multimodal Reasoning by Iterative Perception
  41. NeurIPS 2025VideoRFT: Incentivizing Video Reasoning Capability in MLLMs via Reinforced Fine-Tuning
  42. NeurIPS 2025Vinci: Deep Thinking in Text-to-Image Generation using Unified Model with Reinforcement Learning
  43. NeurIPS 2025VisDiff: SDF-Guided Polygon Generation for Visibility Reconstruction, Characterization and Recognition
  44. NeurIPS 2025Visual Structures Help Visual Reasoning: Addressing the Binding Problem in LVLMs
  45. NeurIPS 2025Visual Thoughts: A Unified Perspective of Understanding Multimodal Chain-of-Thought
  46. NeurIPS 2025VisualQuality-R1: Reasoning-Induced Image Quality Assessment via Reinforcement Learning to Rank
  47. NeurIPS 2025VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning
  48. NeurIPS 2025VLM-R³: Region Recognition, Reasoning, and Refinement for Enhanced Multimodal Chain-of-Thought
  49. NeurIPS 2025VLMs have Tunnel Vision: Evaluating Nonlocal Visual Reasoning in Leading VLMs
  50. NeurIPS 2025VPO: Reasoning Preferences Optimization Based on $\mathcal{V}$-Usable Information

在 Claude、Cursor 里直接问这个趋势

接上 Paperlist MCP 后，写综述、挑方向或准备投稿时，不用离开对话框就能拿到这页的数字：每年各会议的论文数、增长倍数、涨得最快的子方向和代表论文，AI 会按你的问题整理。

可以直接复制这句话问 AI

用 paperlist 查推理模型在 NeurIPS 2023–2026 的论文趋势，哪一年涨得最快？

AI 会调用的工具：get_topic_stats {"topic": "reasoning-models"}

两分钟接入 Paperlist MCP（Claude Code、Cursor、Codex、ChatGPT）

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。

「前沿系统」下的其他主题