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 361 篇，第 5 / 8 页

  1. NeurIPS 2026Neural Garbage Collection: Learning to Forget while Learning to Reason
  2. NeurIPS 2026Neural Spectral Capacity: An Architectural Quantity from Network Specification Alone
  3. NeurIPS 2026Neurosymbolic Learning for Inference-Time Argumentation
  4. NeurIPS 2026Never Stop Learning: Test-Time Reinforcement Learning for Vision-Language-Action Models
  5. NeurIPS 2026Nexus: Same Pretraining Loss, Better Downstream Generalization via Common Minima
  6. NeurIPS 2026NPCBench: A Clinical Apprenticeship Benchmark for Guideline-Constrained Care-Pathway Reasoning in Nasopharyngeal Carcinoma
  7. NeurIPS 2026OmniSimulator: Aligning Small Language Models for Authentic Heterogeneous Behavior Modeling
  8. NeurIPS 2026OmniTraffic: A Controllable Generation Pipeline and Benchmark for Spatio-Temporal Traffic Reasoning
  9. NeurIPS 2026On the Primacy Bias in RLVR Training
  10. NeurIPS 2026On the Token Value Inequality in Efficient Reasoning
  11. NeurIPS 2026One-Layer Transformers Provably Learn In-Context K-Nearest Neighbor Prediction with Chain-of-Thought
  12. NeurIPS 2026OneSearch-V2: The Latent Reasoning Enhanced Self-distillation Generative Search Framework
  13. NeurIPS 2026OpenCoF: Learning to Reason Through Video Generation
  14. NeurIPS 2026OpenMedReason: Scientific Reasoning Supervision for Medical Vision–Language Models
  15. NeurIPS 2026Operads for compositional reasoning in LLMs
  16. NeurIPS 2026OS-Pruner: Pruning Chains-of-Thought of Reasoning Models via Optimal Stopping
  17. NeurIPS 2026Outcome-Based RL Provably Leads Transformers to Reason, but Only With the Right Data
  18. NeurIPS 2026PAC Reasoning: Controlling the Performance Loss for Efficient Reasoning
  19. NeurIPS 2026Pause and Reflect: Conformal Aggregation for Chain-of-Thought Reasoning
  20. NeurIPS 2026Personal Visual Memory from Explicit and Implicit Evidence
  21. NeurIPS 2026PerturbReason: A Knowledge-Grounded Benchmark and Framework for Cell-State–Conditioned Mechanistic Reasoning of Perturbation Effects
  22. NeurIPS 2026Phases of Muon: When Muon Eclipses SignSGD
  23. NeurIPS 2026Plausibility Is Not Prediction: Contrastive Evidence for LLM-Based Cellular Perturbation Reasoning
  24. NeurIPS 2026Point-to-Manifold Geometry: Flexibly Overcoming the Curse of Dimensionality in Neural Computational Units
  25. NeurIPS 2026Position: Robust Reasoning Requires Internal Time, Not Scale Alone
  26. NeurIPS 2026PotARCin: Multi-Dimensional Evaluation of Skill Acquisition in Abstract Reasoning Tasks
  27. NeurIPS 2026Predicting and improving test-time scaling laws via reward tail-guided search
  28. NeurIPS 2026Primal Generation, Dual Judgment: Self-Training from Test-Time Scaling
  29. NeurIPS 2026PRISM-Bench: A Benchmark of Puzzle-Based Visual Tasks with CoT Error Detection
  30. NeurIPS 2026Probabilistic Recursive Reasoning
  31. NeurIPS 2026Probabilistic Tiny Recursive Model
  32. NeurIPS 2026Probing the Trajectories of Reasoning Traces in Large Language Models
  33. NeurIPS 2026Provable Test-Time Scaling for Beam Search in LLM Reasoning
  34. NeurIPS 2026Quantized Reasoning Models Think They Need to Think Longer, but They Do Not
  35. NeurIPS 2026Random Attention Pattern Learning Enables Emergent Capabilities
  36. NeurIPS 2026Reading Between the Dots: Decoding Hidden Computation across Filler Tokens
  37. NeurIPS 2026Reading the Unreadable: Text-Aware Image Super-Resolution Needs Reasoning
  38. NeurIPS 2026Reason to Play: Behavioral and Brain Alignment Between Frontier LRMs and Human Game Learners
  39. NeurIPS 2026Reasoning over Coupled Receptive Fields: Eliminating Subgraph Redundancy at Scale
  40. NeurIPS 2026Reasoning Pathologies in Large Language Models: A Diagnostic Perspective
  41. NeurIPS 2026Reasoning with Sampling: Cutting at Decision Points
  42. NeurIPS 2026Reasoning-Aware IRT: Explainable Evaluation of Test-Time Scaling in Reasoning Models
  43. NeurIPS 2026Reasoning-Trace Collapse: Evaluating the Loss of Explicit Reasoning During Fine-Tuning
  44. NeurIPS 2026ReasoningShield: Safety Moderation over Reasoning Traces of Large Reasoning Models
  45. NeurIPS 2026Rebellious Student: Reversing Teacher Signals for Reasoning Exploration with Self-Distilled RLVR
  46. NeurIPS 2026RECAP: Looking Once Is Not Enough for Vision-Language Reasoning
  47. NeurIPS 2026RefineAny3D: Depth Refinement as Semantic Alignment for Monocular 3D Detection
  48. NeurIPS 2026ReflectMT: Internalizing Reflection for Efficient and High-Quality Machine Translation
  49. NeurIPS 2026Relaxation-Aligned State Control for Test-Time Scaling in Generative Combinatorial Optimization
  50. NeurIPS 2026Relaxed On-Policy Distillation: Selective Credit Allocation for Scaling Reasoning Efficiently

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