MixReasoning: Switching Modes to Think
On the map: LLM reasoning → Test-time compute and efficient thinking
Abstract (arXiv)
Reasoning models enhance performance by tackling problems in a step-by-step manner, decomposing them into sub-problems and exploring long chains of thought before producing an answer. However, applying extended reasoning to every step introduces substantial redundancy, as sub-problems vary widely in difficulty and complexity: a small number of pivotal steps are genuinely challenging and decisive for the final answer, while many others only involve straightforward revisions or simple computations. Therefore, a natural idea is to endow reasoning models with the ability to adaptively respond to this variation, rather than treating all steps with the same level of elaboration. To this end, we propose MixReasoning, a framework that dynamically adjusts the depth of reasoning within a single response. The resulting chain of thought then becomes a mixture of detailed reasoning on difficult steps and concise inference on simpler ones. Experiments on GSM8K, MATH-500, and AIME show that MixReasoning shortens reasoning length and substantially improves efficiency without compromising accuracy.
Key facts. Cited by 3 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 43 of its 61 references that are in the map. First posted on arXiv 7 Oct 2025.
Builds on
43 of 61 references.
Closest ancestors · 25
Same direction, newest first
- 2025Deep Think with Confidence21 Aug
- 2025gpt-oss-120b & gpt-oss-20b Model Card8 Aug
- 2025ConciseHint: Boosting Efficient Reasoning via Continuous Concise Hints during Generation23 Jun
- 2025Wait, We Don't Need to "Wait"! Removing Thinking Tokens Improves Reasoning Efficiency10 Jun
- 2025What makes Reasoning Models Different? Follow the Reasoning Leader for Efficient Decoding8 Jun
- 2025AlphaOne: Reasoning Models Thinking Slow and Fast at Test Time30 May
- 2025VeriThinker: Learning to Verify Makes Reasoning Model Efficient23 May
- 2025AdaptThink: Reasoning Models Can Learn When to Think19 May
- 2025Thinkless: LLM Learns When to Think19 May
- 2025Between Underthinking and Overthinking: An Empirical Study of Reasoning Length and correctness in LLMs30 Apr
- 2025Dynamic Early Exit in Reasoning Models22 Apr
- 2025An Empirical Study of LLM Reasoning Ability Under Strict Output Length Constraint19 Apr
13 more
- 2025THOUGHTTERMINATOR: Benchmarking, Calibrating, and Mitigating Overthinking in Reasoning Models17 Apr
- 2025Efficient Reasoning Models: A Survey15 Apr
- 2025Reasoning Models Can Be Effective Without Thinking14 Apr
- 2025Speculative Thinking: Enhancing Small-Model Reasoning with Large Model Guidance at Inference Time12 Apr
- 2025SpecReason: Fast and Accurate Inference-Time Compute via Speculative Reasoning10 Apr
- 2025L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learning6 Mar
- 2025Scalable Best-of-N Selection for Large Language Models via Self-Certainty25 Feb
- 2025CoT-Valve: Length-Compressible Chain-of-Thought Tuning13 Feb
- 2025Reward-Guided Speculative Decoding for Efficient LLM Reasoning31 Jan
- 2025Thoughts Are All Over the Place: On the Underthinking of o1-Like LLMs30 Jan
- 2025O1-Pruner: Length-Harmonizing Fine-Tuning for O1-Like Reasoning Pruning22 Jan
- 2024Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs30 Dec
- 2024Token-Budget-Aware LLM Reasoning24 Dec
Other references · 18
- 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
- 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,302
- 2021Training Verifiers to Solve Math Word Problemscited by 9,743
- 2025Qwen3 Technical Reportcited by 8,195
- 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,767
- 2023Let's Verify Step by Stepcited by 3,598
- 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,748
- 2022Locating and Editing Factual Associations in GPTcited by 2,609
- 2024LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Codecited by 1,906
- 2018CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledgecited by 1,793
- 2020Transformer Feed-Forward Layers Are Key-Value Memoriescited by 1,307
- 2022Editing Models with Task Arithmeticcited by 1,076
6 more
- 2024Training Large Language Models to Reason in a Continuous Latent Spacecited by 763
- 2021Did Aristotle Use a Laptop? A Question Answering Benchmark with Implicit Reasoning Strategiescited by 736
- 2025Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoningcited by 540
- 2025AIMO-2 Winning Solution: Building State-of-the-Art Mathematical Reasoning Models with OpenMathReasoning datasetcited by 165
- 2024SWIFT: On-the-Fly Self-Speculative Decoding for LLM Inference Accelerationcited by 69
- 2024The Benefits of a Concise Chain of Thought on Problem-Solving in Large Language Modelscited by 62
Built on by
Citing papers per year: 2026: 3.
3 papers.
Similar, not cited
Close in wording, no citation either way
- 2026ThinkRetrieve: Retrieval-Augmented Reasoning Traces for Test-Time Scaling
- 2025The Thinking Spectrum: An Empirical Study of Tunable Reasoning in LLMs through Model Merging
- 2026Efficient Test-Time Scaling via Temporal Reasoning Aggregation
- 2025The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity
- 2026Knowing When Thinking Is Not Enough: Teaching Small Reasoning Models to Reason Beyond Their Parametric Knowledge
- 2026TRACES: Tagging Reasoning Steps for Adaptive Cost-Efficient Early-Stopping
- 2026The Potential of CoT for Reasoning: A Closer Look at Trace Dynamics
- 2026How Much Thinking is Enough? Quantifying and Understanding Redundancy in LLM Reasoning