mlsanityData to 13 Oct 2026

Reasoning on a Budget: A Survey of Adaptive and Controllable Test-Time Compute in LLMs

arXiv 2507.02076 ↗First posted 2 Jul 2025cs.AI cs.LGcited by 31

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

Abstract (arXiv)

Large language models (LLMs) have rapidly progressed into general-purpose agents capable of solving a broad spectrum of tasks. However, current models remain inefficient at reasoning: they apply fixed inference-time compute regardless of task complexity, often overthinking simple problems while underthinking hard ones. This survey presents a comprehensive review of efficient test-time compute (TTC) strategies, which aim to improve the computational efficiency of LLM reasoning. We introduce a two-tiered taxonomy that distinguishes between L1-controllability, methods that operate under fixed compute budgets, and L2-adaptiveness, methods that dynamically scale inference based on input difficulty or model confidence. We benchmark leading proprietary LLMs across diverse datasets, highlighting critical trade-offs between reasoning performance and token usage. Compared to prior surveys on efficient reasoning, our review emphasizes the practical control, adaptability, and scalability of TTC methods. Finally, we discuss emerging trends such as hybrid thinking models and identify key challenges for future work towards making LLMs more computationally efficient, robust, and responsive to user constraints.

Key facts. Cited by 31 papers in the map (data through 13 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 116 of its 127 references that are in the map. First posted on arXiv 2 Jul 2025.

Builds on

116 of 127 references.

Closest ancestors · 54

Same direction, newest first

  1. 2025Adaptive Deep Reasoning: Triggering Deep Thinking When Needed26 May
  2. 2025AdaCtrl: Towards Adaptive and Controllable Reasoning via Difficulty-Aware Budgeting24 May
  3. 2025AdaReasoner: Adaptive Reasoning Enables More Flexible Thinking in Large Language Models22 May
  4. 2025Think Only When You Need with Large Hybrid-Reasoning Models20 May
  5. 2025Let LRMs Break Free from Overthinking via Self-Braking Tuning20 May
  6. 2025Scalable Chain of Thoughts via Elastic Reasoning8 May
  7. 2025Long-Short Chain-of-Thought Mixture Supervised Fine-Tuning Eliciting Efficient Reasoning in Large Language Models6 May
  8. 2025Learning Adaptive Parallel Reasoning with Language Models21 Apr
  9. 2025Concise Reasoning via Reinforcement Learning7 Apr
  10. 2025Retro-Search: Exploring Untaken Paths for Deeper and Efficient Reasoning6 Apr
  11. 2025Harnessing the Reasoning Economy: A Survey of Efficient Reasoning for Large Language Models31 Mar
  12. 2025A Survey of Efficient Reasoning for Large Reasoning Models: Language, Multimodality, and Beyond27 Mar
38 more
  1. 2025Unlocking Efficient Long-to-Short LLM Reasoning with Model Merging26 Mar
  2. 2025Dancing with Critiques: Enhancing LLM Reasoning with Stepwise Natural Language Self-Critique21 Mar
  3. 2025Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models20 Mar
  4. 2025L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learning6 Mar
  5. 2025DAST: Difficulty-Adaptive Slow-Thinking for Large Reasoning Models6 Mar
  6. 2025Sampling-Efficient Test-Time Scaling: Self-Estimating the Best-of-N Sampling in Early Decoding3 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. 2025Meta-Reasoner: Dynamic Guidance for Optimized Inference-time Reasoning in Large Language Models27 Feb
  10. 2025Efficient Test-Time Scaling via Self-Calibration25 Feb
  11. 2025Towards Thinking-Optimal Scaling of Test-Time Compute for LLM Reasoning25 Feb
  12. 2025Dynamic Parallel Tree Search for Efficient LLM Reasoning22 Feb
  13. 2025LightThinker: Thinking Step-by-Step Compression21 Feb
  14. 2025Inner Thinking Transformer: Leveraging Dynamic Depth Scaling to Foster Adaptive Internal Thinking19 Feb
  15. 2025TokenSkip: Controllable Chain-of-Thought Compression in LLMs17 Feb
  16. 2025FastMCTS: A Simple Sampling Strategy for Data Synthesis17 Feb
  17. 2025CoT-Valve: Length-Compressible Chain-of-Thought Tuning13 Feb
  18. 2025Can 1B LLM Surpass 405B LLM? Rethinking Compute-Optimal Test-Time Scaling10 Feb
  19. 2025Demystifying Long Chain-of-Thought Reasoning in LLMs5 Feb
  20. 2025s1: Simple test-time scaling31 Jan
  21. 2025Efficient Reasoning with Hidden Thinking31 Jan
  22. 2025Thoughts Are All Over the Place: On the Underthinking of o1-Like LLMs30 Jan
  23. 2025Think Smarter not Harder: Adaptive Reasoning with Inference Aware Optimization29 Jan
  24. 2025O1-Pruner: Length-Harmonizing Fine-Tuning for O1-Like Reasoning Pruning22 Jan
  25. 2024Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs30 Dec
  26. 2024Inference-Aware Fine-Tuning for Best-of-N Sampling in Large Language Models18 Dec
  27. 2024Compressed Chain of Thought: Efficient Reasoning Through Dense Representations17 Dec
  28. 2024C3oT: Generating Shorter Chain-of-Thought without Compromising Effectiveness16 Dec
  29. 2024Dualformer: Controllable Fast and Slow Thinking by Learning with Randomized Reasoning Traces13 Oct
  30. 2024Learning How Hard to Think: Input-Adaptive Allocation of LM Computation7 Oct
  31. 2024Adaptive Inference-Time Compute: LLMs Can Predict if They Can Do Better, Even Mid-Generation3 Oct
  32. 2024Training Language Models to Self-Correct via Reinforcement Learning19 Sep
  33. 2024Reasoning Aware Self-Consistency: Leveraging Reasoning Paths for Efficient LLM Sampling30 Aug
  34. 2024Path-Consistency with Prefix Enhancement for Efficient Inference in LLMs25 Aug
  35. 2024Make Every Penny Count: Difficulty-Adaptive Self-Consistency for Cost-Efficient Reasoning24 Aug
  36. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters6 Aug
  37. 2024Concise Thoughts: Impact of Output Length on LLM Reasoning and Cost29 Jul
  38. 2024Integrate the Essence and Eliminate the Dross: Fine-Grained Self-Consistency for Free-Form Language Generation2 Jul

50 newest of 54 All 54 (JSON)

Other references · 62

  1. 2017Attention Is All You Needcited by 60,106
  2. 2020Language Models are Few-Shot Learnerscited by 32,320
  3. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,376
  4. 2021Training Verifiers to Solve Math Word Problemscited by 9,916
  5. 2020Scaling Laws for Neural Language Modelscited by 6,494
  6. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,247
  7. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,657
  8. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,270
  9. 2023Let's Verify Step by Stepcited by 3,672
  10. 2025π0.5: a Vision-Language-Action Model with Open-World Generalizationcited by 1,699
  11. 2025Understanding R1-Zero-Like Training: A Critical Perspectivecited by 1,304
  12. 2025Kimi k1.5: Scaling Reinforcement Learning with LLMscited by 1,201
38 more
  1. 2024Large Language Monkeys: Scaling Inference Compute with Repeated Samplingcited by 970
  2. 2025Large Language Diffusion Modelscited by 888
  3. 2024Simple and Effective Masked Diffusion Language Modelscited by 816
  4. 2024Training Large Language Models to Reason in a Continuous Latent Spacecited by 782
  5. 2025SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-trainingcited by 663
  6. 2025Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language Modelscited by 659
  7. 2024A Survey on Large Language Models for Code Generationcited by 656
  8. 2025VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Modelcited by 520
  9. 2025Video-R1: Reinforcing Video Reasoning in MLLMscited by 466
  10. 2024LLaVA-CoT: Let Vision Language Models Reason Step-by-Stepcited by 449
  11. 2022Multimodal Learning with Transformers: A Surveycited by 399
  12. 2025R1-Onevision: Advancing Generalized Multimodal Reasoning through Cross-Modal Formalizationcited by 366
  13. 2024Robotic Control via Embodied Chain-of-Thought Reasoningcited by 343
  14. 2023On the Planning Abilities of Large Language Models : A Critical Investigationcited by 310
  15. 2025R1-VL: Learning to Reason with Multimodal Large Language Models via Step-wise Group Relative Policy Optimizationcited by 299
  16. 2025Chain of Draft: Thinking Faster by Writing Lesscited by 292
  17. 2024Large Language Models for Mathematical Reasoning: Progresses and Challengescited by 284
  18. 2025VAPO: Efficient and Reliable Reinforcement Learning for Advanced Reasoning Taskscited by 283
  19. 2025Training Language Models to Reason Efficientlycited by 245
  20. 2025CODI: Compressing Chain-of-Thought into Continuous Space via Self-Distillationcited by 219
  21. 2023Efficient Large Language Models: A Surveycited by 185
  22. 2025When More is Less: Understanding Chain-of-Thought Length in LLMscited by 182
  23. 2025R1-Zero's "Aha Moment" in Visual Reasoning on a 2B Non-SFT Modelcited by 164
  24. 2024Distilling System 2 into System 1cited by 158
  25. 2024The Impact of Reasoning Step Length on Large Language Modelscited by 152
  26. 2023Progressive-Hint Prompting Improves Reasoning in Large Language Modelscited by 147
  27. 2023LLM4Drive: A Survey of Large Language Models for Autonomous Drivingcited by 136
  28. 2025Optimizing Test-Time Compute via Meta Reinforcement Fine-Tuningcited by 122
  29. 2025Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketchingcited by 116
  30. 2024Fast Best-of-N Decoding via Speculative Rejectioncited by 112
  31. 2025SoftCoT: Soft Chain-of-Thought for Efficient Reasoning with LLMscited by 111
  32. 2023Branch-Solve-Merge Improves Large Language Model Evaluation and Generationcited by 105
  33. 2024Chain of Preference Optimization: Improving Chain-of-Thought Reasoning in LLMscited by 96
  34. 2024Keep the Cost Down: A Review on Methods to Optimize LLM' s KV-Cache Consumptioncited by 96
  35. 2025MATH-Perturb: Benchmarking LLMs' Math Reasoning Abilities against Hard Perturbationscited by 89
  36. 2025Skywork R1V: Pioneering Multimodal Reasoning with Chain-of-Thoughtcited by 69
  37. 2023Evaluating Spatial Understanding of Large Language Modelscited by 65
  38. 2024The Benefits of a Concise Chain of Thought on Problem-Solving in Large Language Modelscited by 62

50 most cited of 62 All 62 (JSON)

Built on by

Cited by year (2026 to 13 Oct)

Citing papers per year: 2025: 6; 2026: 25.

31 papers.

Sort:

By area:

  1. 2025Beyond GPT-5: Making LLMs Cheaper and Better via Performance-Efficiency Optimized Routingcited by 33
  2. 2026Scalable Power Sampling: Unlocking Efficient, Training-Free Reasoning for LLMs via Distribution Sharpeningcited by 29
  3. 2025Towards Concise and Adaptive Thinking in Large Reasoning Models: A Surveycited by 18
  4. 2025ARES: Multimodal Adaptive Reasoning via Difficulty-Aware Token-Level Entropy Shapingcited by 16
  5. 2026Self-Play Only Evolves When Self-Synthetic Pipeline Ensures Learnable Information Gaincited by 16
  6. 2025Real-Time Reasoning Agents in Evolving Environmentscited by 14
  7. 2025From Efficiency to Adaptivity: A Deeper Look at Adaptive Reasoning in Large Language Modelscited by 11
  8. 2026Are Latent Reasoning Models Easily Interpretable?cited by 8
  9. 2026DART: Draft-Agreement Routing for Training-Free Adaptive Thinking Budgets in Hybrid Reasoning Modelscited by 8
  10. 2026CtrlCoT: Dual-Granularity Chain-of-Thought Compression for Controllable Reasoningcited by 7
  11. 2026Ares: Adaptive Reasoning Effort Selection for Efficient LLM Agentscited by 7
  12. 2026TRIAGE: Evaluating Prospective Metacognitive Control in LLMs under Resource Constraintscited by 6
19 more
  1. 2025ARM2: Adaptive Reasoning Model with Vision Understanding and Executable Codecited by 4
  2. 2026Stream-T1: Test-Time Scaling for Streaming Video Generationcited by 4
  3. 2026Does Rank Still Matter? Position Bias When AI Agents Shop on Our Behalfcited by 4
  4. 2026A Relative-Budget Theory for Reinforcement Learning with Verifiable Rewards in Large Language Model Reasoningcited by 3
  5. 2026Learning When to Sample: Confidence-Aware Selective Sampling for Efficient Chain-of-Thought Reasoningcited by 2
  6. 2026Linking Perception, Confidence and Accuracy in MLLMscited by 2
  7. 2026Not All Errors Are Equal: Consequence-Aware Reasoning Compute Allocationcited by 2
  8. 2026AutoCRAT: Within-trajectory Joint Control of Stochasticity and Compute for LLM Reasoningcited by 2
  9. 2026QA-Merging: Query-Adaptive Reasoning via Layer Selective Model Mergingcited by 1
  10. 2026DOVA: Deliberation-First Multi-Agent Orchestration for Autonomous Research Automationcited by 1
  11. 2026Implicit Compression Regularization: Concise Reasoning via Internal Shorter Distributions in RL Post-Trainingcited by 1
  12. 2026ARES: Adaptive Reasoning-Effort Steering for PPA- and Cost-Aware RTL Optimization with LLM Agentscited by 1
  13. 2026Can your AI agent be cheaper? Investigating the effects of task specifications on token spend in agentic coding taskscited by 1
  14. 2026Efficient Reasoning on the Edgecited by 0
  15. 2026STEM Agent: A Self-Adapting, Tool-Enabled, Extensible Architecture for Multi-Protocol AI Agent Systemscited by 0
  16. 2026Detect Early, Escalate Rarely: Anytime Detection of AI-Generated Video from the Compressed Bitstreamcited by 0
  17. 2026Budget-Aware LLM Discovery via Cost-Calibrated Frontier Utilitycited by 0
  18. 2026The Price of Thought: Does Test-Time Reasoning Pay in LLM Trading?cited by 0
  19. 2026FORGE: Form-Optimal Routing of Grounded Evidence for Frozen LLM Agentscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026ThinkBooster: A Unified Framework for Seamless Test-Time Scaling of LLM Reasoning
  2. 2025The Art of Scaling Test-Time Compute for Large Language Models
  3. 2025Inference-Time Computations for LLM Reasoning and Planning: A Benchmark and Insights
  4. 2024Inference Scaling vs Reasoning: An Empirical Analysis of Compute-Optimal LLM Problem-Solving
  5. 2025A Survey of Frontiers in LLM Reasoning: Inference Scaling, Learning to Reason, and Agentic Systems
  6. 2025Prompting Test-Time Scaling Is A Strong LLM Reasoning Data Augmentation
  7. 2025Less is More: Improving LLM Reasoning with Minimal Test-Time Intervention
  8. 2025A Survey of Scaling in Large Language Model Reasoning