mlsanityData to 7 Oct 2026

Does Your Reasoning Model Implicitly Know When to Stop Thinking?

arXiv 2602.08354 ↗First posted 9 Feb 2026cs.AIcited by 17

On the map: LLM reasoning → Test-time compute and efficient thinking

Abstract (arXiv)

Recent advancements in large reasoning models (LRMs) have greatly improved their capabilities on complex reasoning tasks through Long Chains of Thought (CoTs). However, this approach often results in substantial redundancy, impairing computational efficiency and causing significant delays in real-time applications. Recent studies show that longer reasoning chains are frequently uncorrelated with correctness and can even be detrimental to accuracy. In a further in-depth analysis of this phenomenon, we surprisingly uncover and empirically verify that LRMs implicitly know the appropriate time to stop thinking, while this capability is obscured by current sampling paradigms. Motivated by this, we introduce SAGE (Self-Aware Guided Efficient Reasoning), a novel sampling paradigm that unleashes this efficient reasoning potential. Furthermore, integrating SAGE as mixed sampling into group-based reinforcement learning (SAGE-RL) enables SAGE-RL to effectively incorporate SAGE-discovered efficient reasoning patterns into standard pass@1 inference, markedly enhancing both the reasoning accuracy and efficiency of LRMs across multiple challenging mathematical benchmarks.

Key facts. Cited by 17 papers in the map (data through 7 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 70 of its 115 references that are in the map. First posted on arXiv 9 Feb 2026.

Builds on

70 of 115 references.

Closest ancestors · 29

Same direction, newest first

  1. 2025SpecExit: Accelerating Large Reasoning Model via Speculative Exit29 Sep
  2. 2025Promoting Efficient Reasoning with Verifiable Stepwise Reward14 Aug
  3. 2025Optimizing Length Compression in Large Reasoning Models17 Jun
  4. 2025CyclicReflex: Improving Reasoning Models via Cyclical Reflection Token Scheduling4 Jun
  5. 2025AdaCtrl: Towards Adaptive and Controllable Reasoning via Difficulty-Aware Budgeting24 May
  6. 2025Don't Overthink it. Preferring Shorter Thinking Chains for Improved LLM Reasoning23 May
  7. 2025Optimizing Anytime Reasoning via Budget Relative Policy Optimization19 May
  8. 2025AdaptThink: Reasoning Models Can Learn When to Think19 May
  9. 2025ConCISE: Confidence-guided Compression in Step-by-step Efficient Reasoning8 May
  10. 2025ShorterBetter: Guiding Reasoning Models to Find Optimal Inference Length for Efficient Reasoning30 Apr
  11. 2025Dynamic Early Exit in Reasoning Models22 Apr
  12. 2025Thought Manipulation: External Thought Can Be Efficient for Large Reasoning Models18 Apr
17 more
  1. 2025Reasoning Models Can Be Effective Without Thinking14 Apr
  2. 2025SEAL: Steerable Reasoning Calibration of Large Language Models for Free7 Apr
  3. 2025ThinkPrune: Pruning Long Chain-of-Thought of LLMs via Reinforcement Learning2 Apr
  4. 2025Inference-Time Scaling for Complex Tasks: Where We Stand and What Lies Ahead31 Mar
  5. 2025L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learning6 Mar
  6. 2025DAST: Difficulty-Adaptive Slow-Thinking for Large Reasoning Models6 Mar
  7. 2025How Well do LLMs Compress Their Own Chain-of-Thought? A Token Complexity Approach3 Mar
  8. 2025Self-Training Elicits Concise Reasoning in Large Language Models27 Feb
  9. 2025TokenSkip: Controllable Chain-of-Thought Compression in LLMs17 Feb
  10. 2025CoT-Valve: Length-Compressible Chain-of-Thought Tuning13 Feb
  11. 2025Demystifying Long Chain-of-Thought Reasoning in LLMs5 Feb
  12. 2025Reward-Guided Speculative Decoding for Efficient LLM Reasoning31 Jan
  13. 2025O1-Pruner: Length-Harmonizing Fine-Tuning for O1-Like Reasoning Pruning22 Jan
  14. 2024Token-Budget-Aware LLM Reasoning24 Dec
  15. 2024Adaptive Inference-Time Compute: LLMs Can Predict if They Can Do Better, Even Mid-Generation3 Oct
  16. 2024Escape Sky-high Cost: Early-stopping Self-Consistency for Multi-step Reasoning19 Jan
  17. 2023Let's Sample Step by Step: Adaptive-Consistency for Efficient Reasoning and Coding with LLMs19 May

Other references · 41

  1. 2017Proximal Policy Optimization Algorithmscited by 14,441
  2. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 8,851
  3. 2025Qwen3 Technical Reportcited by 8,195
  4. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  5. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,146
  6. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
  7. 2023Let's Verify Step by Stepcited by 3,598
  8. 2025DAPO: An Open-Source LLM Reinforcement Learning System at Scalecited by 2,678
  9. 2024HybridFlow: A Flexible and Efficient RLHF Frameworkcited by 1,980
  10. 2025Kimi k1.5: Scaling Reinforcement Learning with LLMscited by 1,193
  11. 2024Tulu 3: Pushing Frontiers in Open Language Model Post-Trainingcited by 988
  12. 2024SimPO: Simple Preference Optimization with a Reference-Free Rewardcited by 951
29 more
  1. 2025Group Sequence Policy Optimizationcited by 788
  2. 2025SimpleRL-Zoo: Investigating and Taming Zero Reinforcement Learning for Open Base Models in the Wildcited by 505
  3. 2025Chain of Draft: Thinking Faster by Writing Lesscited by 290
  4. 2025Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMscited by 276
  5. 2025Training Language Models to Reason Efficientlycited by 242
  6. 2023ReMax: A Simple, Effective, and Efficient Reinforcement Learning Method for Aligning Large Language Modelscited by 184
  7. 2025Light-R1: Curriculum SFT, DPO and RL for Long COT from Scratch and Beyondcited by 175
  8. 2024Distilling System 2 into System 1cited by 152
  9. 2025Optimizing Test-Time Compute via Meta Reinforcement Fine-Tuningcited by 121
  10. 2025Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketchingcited by 116
  11. 2025Act Only When It Pays: Efficient Reinforcement Learning for LLM Reasoning via Selective Rolloutscited by 108
  12. 2024On Designing Effective RL Reward at Training Time for LLM Reasoningcited by 87
  13. 2025S-GRPO: Early Exit via Reinforcement Learning in Reasoning Modelscited by 87
  14. 2025GRPO-LEAD: A Difficulty-Aware Reinforcement Learning Approach for Concise Mathematical Reasoning in Language Modelscited by 81
  15. 2025Sample More to Think Less: Group Filtered Policy Optimization for Concise Reasoningcited by 75
  16. 2023HIPTrack: Visual Tracking with Historical Promptscited by 71
  17. 2024Can Language Models Learn to Skip Steps?cited by 59
  18. 2025Stepwise Perplexity-Guided Refinement for Efficient Chain-of-Thought Reasoning in Large Language Modelscited by 51
  19. 2025ARM: Adaptive Reasoning Modelcited by 38
  20. 2022SparseTT: Visual Tracking with Sparse Transformerscited by 36
  21. 2026Your Group-Relative Advantage Is Biasedcited by 32
  22. 2025FastCuRL: Curriculum Reinforcement Learning with Stage-wise Context Scaling for Efficient Training R1-like Reasoning Modelscited by 26
  23. 2025Adaptive Batch-Wise Sample Scheduling for Direct Preference Optimizationcited by 16
  24. 2025AdaMCoT: Rethinking Cross-Lingual Factual Reasoning through Adaptive Multilingual Chain-of-Thoughtcited by 14
  25. 2025Interpretable Unsupervised Joint Denoising and Enhancement for Real-World low-light Scenarioscited by 11
  26. 2025LLMBoost: Make Large Language Models Stronger with Boostingcited by 8
  27. 2026PRPO: Aligning Process Reward with Outcome Reward in Policy Optimizationcited by 7
  28. 2025SeeDNorm: Self-Rescaled Dynamic Normalizationcited by 6
  29. 2026Anchored Policy Optimization: Mitigating Exploration Collapse Via Support-Constrained Rectificationcited by 5

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 17.

17 papers.

Sort:

By area:

  1. 2026Heterogeneous Agent Collaborative Reinforcement Learningcited by 13
  2. 2026MedSynapse-V: Bridging Visual Perception and Clinical Intuition via Latent Memory Evolutioncited by 12
  3. 2026Counterfactual Credit Policy Optimization for Multi-Agent Collaborationcited by 10
  4. 2026Adaptive Robust Estimator for Multi-Agent Reinforcement Learningcited by 2
  5. 2026Policy Improvement Reinforcement Learningcited by 2
  6. 2026Uni-MDTrack: Learning Decoupled Memory and Dynamic States for Parameter-Efficient Visual Tracking in All Modalitycited by 1
  7. 2026CopT: Contrastive On-Policy Thinking with Continuous Spaces for General and Agentic Reasoningcited by 1
  8. 2026VistaHop: Benchmarking Long-Horizon Visual DeepSearchcited by 1
  9. 2026Dynamic Rollout Editing for Reducing Overthinking in RL-Trained Reasoning Modelscited by 1
  10. 2026SpatialCLI: Learning to Reason With Spatial Tools, Then Without Themcited by 1
  11. 2026Placing Puzzle Pieces Where They Matter: A Question Augmentation Framework for Reinforcement Learningcited by 0
  12. 2026EchoStyle: Unlocking High-Fidelity Video Stylization with Reverse Data Synthesiscited by 0
5 more
  1. 2026Models as Tools: An Agentic Coordination Framework for Unified Multimodal Visual Trackingcited by 0
  2. 2026Deferred Exposure of Future Trajectories for Verifiable Reasoning in Autonomous Driving VLMscited by 0
  3. 2026Constraint-First Reasoning: A Training-Free Protocol for Exploiting Answer-Space Constraints in Mathematical Problem Solvingcited by 0
  4. 2026WDL-OPD: Weak-Driven On-Policy Distillation via Mixture-Constrained Co-Trainingcited by 0
  5. 2026Learning to Stop without Learning to Stop: Self-Supervised Confidence Training Improves Reasoning Efficiencycited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Large Reasoning Models are not thinking straight: on the unreliability of thinking trajectories
  2. 2025How Far Are We from Optimal Reasoning Efficiency?
  3. 2025Lost at the Beginning of Reasoning
  4. 2026Towards a Mechanistic Understanding of Large Reasoning Models: A Survey of Training, Inference, and Failures
  5. 2025Don't Think Longer, Think Wisely: Optimizing Thinking Dynamics for Large Reasoning Models
  6. 2025OThink-R1: Intrinsic Fast/Slow Thinking Mode Switching for Over-Reasoning Mitigation
  7. 2025Not All Thoughts are Generated Equal: Efficient LLM Reasoning via Multi-Turn Reinforcement Learning
  8. 2025Exploring and Exploiting the Inherent Efficiency within Large Reasoning Models for Self-Guided Efficiency Enhancement