Efficient Language Model Reasoning

Follow this topic Ask about this topic

This field asks how language models can solve tasks such as math, coding, science questions, and multi-hop retrieval without spending unnecessary computation. Common approaches compress or prune chain-of-thought, stop generation when answers stabilize, and adapt reasoning budgets to task difficulty; others use reinforcement learning, distillation, search, parallel sampling, or smaller-model routing. Newer work also explores latent reasoning and memory or KV-cache management to reduce the cost of generating and retaining explicit thoughts. The central tension is that extra reasoning can help on difficult problems but may waste resources, introduce errors, or even hurt accuracy, so methods are judged by both quality and cost on benchmarks such as GSM8K, MATH, AIME, and GPQA.

Latest papers

373 papers89 papers in the last four weeks, up 218% on the four weeks before. 0.5% of all new papers.

ThinkJev: One Thought, Every Budget for Calibrated Typed Decisions
Oct 27, 2026Xiaolu Ding, Kai Yang, Dongdong LiuLLM Decision-Making Efficient Language Model Reasoning
Fujian Bosssoft Co., Ltd., Fuzhou, China
Reasoning Patterns for Efficient AI Coding: An Empirical Study
Oct 26, 2026Zhe Yin, Yuling Shi, Xiaodong Gu +1Efficient Language Model Reasoning Large Reasoning Models
Shanghai Jiao Tong University, China
Efficient Proof Generation for Lean Statements with Multiple Candidate Answers
Oct 25, 2026Pengzhen Lin, Ke Zhang, Maziar Raissi +1Automated Theorem Proving LLM Mathematical Reasoning
University of California, Riverside
Improving the Chain-of-Thought Compressibility of Pruned Reasoning Models via Long--Short Contrastive Pruning
Oct 25, 2026Tao Jing, Chen Kang, Feifei Zhai +2Efficient Language Model Reasoning LLM Pruning
More: Mode-Optimized Reasoning with Self-Adaptive Effort for Self-Evolving Agents
Oct 24, 2026Dianxing Shi, Dingjie Fu, Tianyu Lei +1Adaptive Reasoning in Language Models Efficient Language Model Reasoning
Beihang University · Huazhong University of Technology
Correct but Unselectable: The Hidden Interface Tax in Multi-Candidate Reasoning
Oct 23, 2026Qianli Ma, Zhiqing Tang, Fanshuai Meng +3Efficient Language Model Reasoning Test-Time Scaling
Faculty of Arts and Science, Beijing Normal University, Zhuhai 519087, China · Institute of Artificial Intelligence and Future Networks, Beijing Normal University, Zhuhai 519087, Guangdong, China · School of Artificial Intelligence, Beijing Normal University, Beijing 100875, China +1
Direct Pass@ $k$ Optimization for Thought Pruning: Balancing Quality and Diversity in Test-Time Reasoning
Oct 22, 2026Mahdi Salmani, Adel Javanmard, Devansh Gupta +1Efficient Language Model Reasoning LLM Pruning
University of Southern California
Same Beliefs, Different Policies: Learning Reusable Representations for Early Stopping in LLM Reasoning
Oct 21, 2026Jingwei Ni, Dmitry Abulkhanov, Vladislav Smirnov +5Early Stopping for Language Model Reasoning Early Stopping
ETH Zürich · MBZUAI · University of Zürich +1
Route-and-Generate: KL-Regularized Router Improvement for Reasoning Routing in LLMs
Oct 21, 2026Tien-Phat Nguyen, Truong Nguyen, Thanh-Thien Le +4Adaptive Reasoning in Language Models Efficient Language Model Reasoning
Hanoi University of Science and Technology, Vietnam · Monash University, Australia · New York University, USA +1
ReAPO: Retry-Aware Policy Optimization for Efficient Exploration in LLM Reasoning
Oct 20, 2026Nghi Dao, Calder Morton-Ferguson, Noah Golowich +1RL for Language Model Reasoning Efficient Language Model Reasoning
Columbia University · Yale University · UT Austin
Adaptive Stopping of Streamed Reasoning for Language Model Agents
Oct 20, 2026JunHyeok Oh, Byung-Jun LeeEarly Stopping for Language Model Reasoning Adaptive Reasoning in Language Models
Korea University
When Can LLM Reasoning Reliably Reuse Intermediate Results?
Oct 20, 2026Deyuan Liu, Zecheng Wang, Xuanyi Liu +5LLM Reasoning Efficient Language Model Reasoning
Westlake University · Wechat AI · Peking University +3
On the Fragility of Tasks under Composition: Implicit Budgeting in Language Models
Oct 19, 2026Chenpan Li, Thies Brockmoeller, Stefanie Jegelka +2Compositional Reasoning Efficient Language Model Reasoning
TU Munich · MIT CSAIL
PORA: Learning When More Reasoning Pays Off for LLM Agents
Oct 19, 2026Bo Yan, Qi Zhang, Zhen Tan +2Adaptive Reasoning in Language Models Efficient Language Model Reasoning
University of Central Florida · Independent Researcher · Stevens Institute of Technology +1
SOMA: Continuous Monitoring and Intervention for System Two Reasoning with System One Models
Oct 19, 2026Ruizhi Qian, Lei Shi, Tianyu Hu +3Adaptive Reasoning in Language Models Metacognition in Language Models
University of Central Florida
BOSS: Token-Budgeted Online Sampling and Selection for LLM Reasoning
Oct 19, 2026Yupeng Chang, Yahan Li, Chenlu Guo +3Efficient Language Model Reasoning Inference-Time Compute Allocation
School of Artificial Intelligence, Jilin University · Xingchen AGI Lab, China Telecom Artificial Intelligence (Beijing) Co., Ltd., China · Bilibili Inc., China +3
GraphEvoResearcher: Graph-Aware Training-Free Self-Reflective Reasoning for Cost-Bounded Multi-Hop Question Answering
Oct 19, 2026Suxing LiuMulti-Hop QA LLM Self-Refinement
Department of Computing, Universiti Sains Malaysia
GLOBE: Geometry-Led Online Behavioral Early Exit via Unit-Sphere Trajectories
Oct 19, 2026Xinyue Peng, Zhenlin Liang, Meng Cai +1Efficient Language Model Reasoning CoT Reasoning
Moore Threads
ABACUS: Program-Guided Multistep Reasoning over Frozen Judgment Models
Oct 19, 2026Yuhao Wang, Zelin ZangStructured Reasoning in Language Models Efficient Language Model Reasoning
Westlake University · Hong Kong Institute of Science & Innovation, CAS · Tsientang Institute for Advanced Study
CALM: Consistency-Aware Logit Masking in Decoding Space for Mitigating Overthinking of Large Reasoning Models
Oct 19, 2026Shengjing Liu, Feiyang Li, Yuxuan Yang +2Overthinking in Language Models Efficient Language Model Reasoning
College of Computer Science and Software Engineering, Shenzhen University
Learning to Reason Efficiently with Interventional Step Credit
Oct 19, 2026Guanda Li, Xuefeng Bai, Yulong Chen +3Efficient Language Model Reasoning Step-Level Credit Assignment in RL
Harbin Institute of Technology, Shenzhen · University of Aberdeen · University of Cambridge
Truncated Reasoning Under a Budget: Evaluating System-1 Models as Decision-Makers
Oct 19, 2026Zhongyin Wang, Zhaoqian Xue, Junting Mu +4LLM Decision-Making Efficient Language Model Reasoning
University of Illinois Urbana-Champaign · University of Arizona · University of California, Irvine +3
Calibration-Free Early Stopping for Chain-of-Thought Reasoning via E-processes
Oct 18, 2026Dong-Jae Lee, Gyojin Han, Junmo KimEarly Stopping for Language Model Reasoning Adaptive Reasoning in Language Models
KAIST
SteadyThought: Mitigating LLM Under-Thinking via Thought-Level Preference Optimization
Oct 18, 2026Zhenyue Peng, Lin Yang, Jie Wang +7Efficient Language Model Reasoning Adaptive Reasoning in Language Models
Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei 230031, China · University of Science and Technology of China, Hefei 230026, China · Shanghai Artificial Intelligence Laboratory, Shanghai, China +3
Reasoning with Context Management and External Memory
Oct 16, 2026Man-Ling Sung, William Hartmann, Jiangying ZhouLLM Context Management Efficient Language Model Reasoning
RTX BBN Technologies, Cambridge MA, USA
Think More or Read More? A Token-Budget Theory of Reasoning with Shared Context
Oct 16, 2026Manh Nguyen, Sunil Gupta, Hung LeEfficient Language Model Reasoning Multi-Agent Debate
Applied Artificial Intelligence Initiative, Deakin University, Australia
Not All Test-Time Compute Is Evidence: Model-Specific Bias in LLM Reasoning
Oct 16, 2026Ming Shen, Tianjun YaoTest-Time Scaling Efficient Language Model Reasoning
Shanghai Jiao Tong University · Shenzhen Loop Area Institute (SLAI)
Sep-Latent: Separator Tokens as Natural Compression Anchors for Implicit Reasoning
Oct 16, 2026Yinlong Xu, Jianfeng Huang, Zhengdong Zhou +5Efficient Language Model Reasoning Implicit Reasoning in Language Models
College of Computer Science and Technology, Zhejiang University, Hangzhou, China · Alibaba Group, Hangzhou, China
Uncertainty-Aware Budget Allocation for Efficient LLM Tree Search
Oct 16, 2026Jianguo Huang, Zhiyi Lyu, Fuxiang Zhang +3Efficient Language Model Reasoning Inference-Time Search
Nanyang Technological University, Singapore
CLUE: Uncovering Social Context for Pragmatic Understanding
Oct 15, 2026Chengfei Wu, Zhaoqing Wu, Dan GoldwasserPragmatic Reasoning in Language Models Social Reasoning
Purdue University

Oct 27, 2026J/K move · Enter open · S save

ThinkJev: One Thought, Every Budget for Calibrated Typed Decisions

Xiaolu Ding, Kai Yang, Dongdong Liu

Fujian Bosssoft Co., Ltd., Fuzhou, China

LLM Decision-Making Efficient Language Model Reasoning Selective Prediction Early Stopping for Language Model Reasoning

Open paper AskPDF arXiv

Many uses of language models end in a typed decision: a choice among a closed set of options, whose probabilities code can act on. Reading those probabilities from one forward pass is fast but often wrong on decisions that take several steps, such as arithmetic, counting, and date comparison. Letting the model think first fixes most of these errors, but costs seconds per decision instead of milliseconds. To pay for thinking only where it helps, ThinkJev builds a fast and a slow system from one open hybrid-thinking model that can answer directly or after a thought. System 1 reads the model's probabilities for the option letters when it answers directly; System 2 reads the same letters after the model has thought on the same cached prompt, so escalating repeats no work and both systems give probabilities over the same options. Because the letters can be read after any number of thought tokens, one sampled thought also gives the answer under every thinking budget. Without training, ThinkJev lets System 1 answer alone only where a statistical test on labeled dev data guarantees that its accepted answers stay under a target error rate. At a 5% target, Qwen3.5-27B answers 24.5% of 8359 test decisions this way with a realized error of 3.9% (Qwen3.5-4B and 9B: 6.5% and 0.0%). For the rest, stopping each thought as soon as its answer is confident reaches 81.6% accuracy, against 82.7% for always thinking, with 72% of the thought tokens. Choosing when to think from System 1's confidence alone saves less, because the decisions System 1 is least sure of are often not the ones thinking fixes. On the public items of JevBench, an independent benchmark of typed-decision services, Qwen3.5-27B thinks on 29.4% and is not distinguishable in accuracy from the two thinking APIs listed (exploratory). Code and data generators are available at https://github.com/ding-xiaolu/ThinkJev.