{"total":16,"items":[{"citing_arxiv_id":"2606.20295","ref_index":86,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Token-Operations-Oriented Inference Optimization Techniques for Large Models","primary_cat":"cs.SE","submitted_at":"2026-06-18T14:33:09+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":3.0,"formal_verification":"none","one_line_summary":"The paper introduces a four-layer technical architecture for token-operations-oriented inference optimization in large models and reviews key technologies and industry status at each layer.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"reasoning; it therefore suppresses such tokens during inference to reduce redundant reasoning without requiring retraining [85]. Abstract-CoT takes a different approach by constructing a discrete latent reasoning mechanism in which models reason using abstract symbols rather than natural language, reducing reasoning-token usage by 11.6x while maintaining comparable performance [86]. CoThink adopts a dual-model collaboration paradigm, in which an instruction-following model first generates a high-level solution outline and a reasoning model subsequently expands it into a detailed solution, achieving a 22.3% reduction in token consumption across three major benchmark datasets [87]. The NAT framework incorporates token budget as a primary optimization objective during reinforcement learning and"},{"citing_arxiv_id":"2606.11470","ref_index":87,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"The Periodic Table of LLM Reasoning: A Structured Survey of Reasoning Paradigms, Methods, and Failure Modes","primary_cat":"cs.CL","submitted_at":"2026-06-09T21:59:37+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":4.0,"formal_verification":"none","one_line_summary":"A literature survey that introduces a taxonomy for LLM reasoning paradigms, analyzes methodological trends, and synthesizes failure modes from over 300 papers.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.03800","ref_index":22,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Trading Human Curation for Synthetic Augmentation in RLVR","primary_cat":"cs.LG","submitted_at":"2026-06-02T15:48:28+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":4.0,"formal_verification":"none","one_line_summary":"Gated synthetic augmentations can substitute for additional human-authored RLVR tasks at a cost-adjusted trade rate of 1.4x-11.6x while retaining held-out generalization on ten benchmarks spanning code, instruction following, reasoning, and agentic function calling.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.02282","ref_index":33,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"POIROT: Interrogating Agents for Failure Detection in Multi-Agent Systems","primary_cat":"cs.AI","submitted_at":"2026-06-01T14:05:35+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"POIROT protocol repurposes agents in LLM multi-agent systems as an internal diagnostic layer for failure detection, outperforming single-LLM evaluators with gains that increase with complexity, agent count, and fault types.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.06165","ref_index":107,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Post Reasoning: Improving the Performance of Non-Thinking Models at No Cost","primary_cat":"cs.AI","submitted_at":"2026-05-07T12:51:49+00:00","verdict":"CONDITIONAL","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Post-Reasoning boosts LLM accuracy by reversing the usual answer-after-reasoning order, delivering mean relative gains of 17.37% across 117 model-benchmark pairs with zero extra cost.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.21764","ref_index":12,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Thinking with Reasoning Skills: Fewer Tokens, More Accuracy","primary_cat":"cs.AI","submitted_at":"2026-04-23T15:12:58+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Distilling and retrieving reusable reasoning skills lets LLMs solve coding and math problems with fewer tokens and higher accuracy.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.23926","ref_index":10,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"How Much Thinking is Enough? Quantifying and Understanding Redundancy in LLM Reasoning","primary_cat":"cs.AI","submitted_at":"2026-04-21T05:32:02+00:00","verdict":"CONDITIONAL","verdict_confidence":"MODERATE","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Across four frontier reasoning models, 61–93% of correct chain-of-thought steps are redundant, and this over-thinking is provably optimal under any length-agnostic outcome reward.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.08290","ref_index":6,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Tokalator: A Context Engineering Toolkit for Artificial Intelligence Coding Assistants","primary_cat":"cs.SE","submitted_at":"2026-04-09T14:27:02+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":4.0,"formal_verification":"none","one_line_summary":"Tokalator is a toolkit with VS Code extension, calculators, and community resources to monitor and optimize token usage in AI coding environments.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.03679","ref_index":11,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"LightThinker++: From Reasoning Compression to Memory Management","primary_cat":"cs.CL","submitted_at":"2026-04-04T10:46:09+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"LightThinker++ adds explicit adaptive memory management and a trajectory synthesis pipeline to LLM reasoning, cutting peak token use by ~70% while gaining accuracy in standard and long-horizon agent tasks.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"5 technical report.CoRR, abs/2412.15115, 2024. doi: 10.48550/ARXIV.2412.15115. URLhttps://doi.org/10.48550/arXiv.2412.15115. [10] Tingxu Han, Zhenting Wang, Chunrong Fang, Shiyu Zhao, Shiqing Ma, and Zhenyu Chen. Token-budget-aware LLM reasoning. CoRR, abs/2412.18547, 2024. doi: 10.48550/ARXIV.2412.18547. URLhttps://doi.org/10. 48550/arXiv.2412.18547. [11] Mengru Ding, Hanmeng Liu, Zhizhang Fu, Jian Song, Wenbo Xie, and Yue Zhang. Break the chain: Large language models can be shortcut reasoners.CoRR, abs/2406.06580, 2024. doi: 10.48550/ARXIV.2406.06580. URL https://doi.org/10.48550/arXiv.2406.06580. [12] Sania Nayab, Giulio Rossolini, Giorgio C. Buttazzo, Nicolamaria Manes, and Fabrizio Giacomelli. Concise thoughts:"},{"citing_arxiv_id":"2509.21743","ref_index":10,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Retrieval-of-Thought: Efficient Reasoning via Reusing Thoughts","primary_cat":"cs.AI","submitted_at":"2025-09-26T01:17:35+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Retrieval-of-Thought organizes prior reasoning into a thought graph for retrieval and reward-guided recombination, reducing output tokens by up to 40% and latency by 82% while preserving accuracy on reasoning benchmarks.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2509.14004","ref_index":4,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Early Stopping Chain-of-thoughts in Large Language Models","primary_cat":"cs.CL","submitted_at":"2025-09-17T14:14:05+00:00","verdict":"CONDITIONAL","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"ES-CoT shortens LLM chain-of-thought generation by tracking runs of identical step answers after linguistic markers, cutting tokens 16% on average while keeping accuracy comparable to full CoT across six datasets and three models.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2509.05489","ref_index":17,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Self-Aligned Reward: Towards Effective and Efficient Reasoners","primary_cat":"cs.LG","submitted_at":"2025-09-05T20:39:43+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Self-aligned reward uses relative perplexity differences to encourage concise, query-specific reasoning in LLMs, yielding 4% accuracy gains and 30% lower inference cost when added to PPO or GRPO.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2505.15858","ref_index":11,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Search-Based Multi-Trajectory Refinement for Safe C-to-Rust Translation with Large Language Models","primary_cat":"cs.PL","submitted_at":"2025-05-21T01:26:23+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"LAC2R uses MCTS to systematically explore multiple LLM refinement trajectories for C-to-Rust translation and reports superior safety and correctness on small-scale benchmarks.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2503.16419","ref_index":58,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models","primary_cat":"cs.CL","submitted_at":"2025-03-20T17:59:38+00:00","verdict":"ACCEPT","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"A survey organizing techniques to achieve efficient reasoning in LLMs by shortening chain-of-thought outputs.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"computation-sensitive real-world scenarios, such as real-time autonomous driving systems, interactive conversational assistants, precision robotic control tasks, and large-scale online search engines. Efficient reasoning, particularly the reduction of reasoning length, offers significant benefits in such regards, providing direct cost reduction and improved feasibility for real-world deployments. Recently, numerous studies [58, 60, 127, 130, 217] have explored ways to develop more concise reasoning paths, making efficient reasoning a rapidly evolving research area. In this paper, we present the first structured survey systematically exploring the progress in efficient reasoning for LLMs. As illustrated in Figure 2, we categorize existing work into three key directions: (1) Model-based efficient reasoning , which focuses on optimizing full-length reasoning models"},{"citing_arxiv_id":"2503.09567","ref_index":236,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Towards Reasoning Era: A Survey of Long Chain-of-Thought for Reasoning Large Language Models","primary_cat":"cs.AI","submitted_at":"2025-03-12T17:35:03+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"The paper unifies perspectives on Long CoT in reasoning LLMs by introducing a taxonomy, detailing characteristics of deep reasoning and reflection, and discussing emergence phenomena and future directions.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":",CoT [836], Reflection of thought [1118], MathPrompter [303], CLP [617], AutoCAP [1070], Plan-Search [765], NaturalProgram [460], CodeI/O [401], Wang et al. [823],etc. Structured LanguageDeep Reasoninge.g.,PoT [100], CoC [389], Brain [107], SIaM [978], ENVISIONS [897], SKIntern [443], QuaSAR [634],TinyGSM [464], MathDivide [687], Payoungkhamdee et al. [591], GPT-f [605], STP [158],etc. Latent SpaceDeep Reasoninge.g.,Quiet-STaR [1013], PlaningToken [810], Coconut [236], RecurrentBlock[204], MuSR [684], SERT[1069], Heima [662], LTMs [356], ITT [109], Deng et al. [151],etc. Deep ReasoningLearning (§4.2) Deep ReasoningImitation e.g.,GSM8K [141], AceMath [500], DART-Math [738], O1-Journey-P2[299], STILL-2 [550], LIMO[967], s1 [560], RedSTaR [902], Fine-tune-CoT [256], CoT-Collection [352], FastMCTS [410],etc. Deep ReasoningSelf-Learninge."},{"citing_arxiv_id":"2412.21187","ref_index":279,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs","primary_cat":"cs.CL","submitted_at":"2024-12-30T18:55:12+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"o1-like models overthink easy tasks; self-training reduces compute use without accuracy loss on GSM8K, MATH500, GPQA, and AIME.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null}],"limit":50,"offset":0}