mlsanityData to 7 Oct 2026

Reasoning for agents, planning and code

2,143 papers14 this weekLow confidence in this grouping: the papers may belong to more than one topic, or the group may changeLow confidence in this grouping: the papers may belong to more than one topic, or the group may change · About

Keywords: reasoning, multi-agent, agents, llm, code

Key facts. Reasoning for agents, planning and code: 2,143 papers on arXiv since 2022 (data through 7 Oct 2026). 149 of its 2,143 mapped papers (7%) cite “OpenAI o1 System Card” (2024, arXiv 2412.16720). 111 of its 2,143 mapped papers (5%) cite “LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code” (2024, arXiv 2403.07974). 81 new papers 17 Sep–7 Oct.

What’s new

17 Sep – 7 Oct 2026

81 new papers, 17 Sep – 7 Oct; 14 this week.

  1. This week · 14
  2. BRANCH: Bypassing Multi-Scanner AI Guardrails
  3. From Pareto to Preference: Personalized Test-Time Scaling via Amortized Agentic Policy Discovery also fits: Reinforcement learning for reasoning
  4. System Switch: When Should a Fast Decision Model Stop and Think?
  5. InsClaimBench: Benchmarking Insurance Claim Adjudication Across the Decision Chain
  6. LLM-Enabled UAV Dispatch: A System-Level Survey and Taxonomy
  7. AdaGuard: Enhancing Safety and Policy Compliance with Reasoning-Enabled LLM-As-A-Judge Guardrails
  8. Sequential Probabilistic Uncertainty Estimation for Parallel Multi-Agent Reasoning Systems
  9. Confidence Reasoning Graphs: Structured Confidence Estimation for LLM Agents
  10. DHCG: Dynamic Construction of Hierarchical Collaboration Graphs for LLM-Based Multi-Agent Reasoning
  11. OOPMAS: Object-Oriented Multi-Agent Systems for Query-Level Workflow Generation
All 81
  1. CLIFT: Conformal Self-Verification for Web Agent Training and Test-Time Scaling
  2. Back to the Future: Rethinking EDA Infrastructure for Agentic Systems in Chip Design Verification
  3. Better Call Reward: Reward Hacking as Strategic Abstention in Legal Reasoning Models also fits: Reasoning benchmarks and evaluation
  4. Correct Verdicts, Flawed Reasoning: Structured Auditing of LLM-based Vulnerability Reasoning
  5. Earlier · 67
  6. StateWise: Diagnosing and Repairing Persistent Operational State Before Agent Actions
  7. Safe Context Switching for Agents in the Wild: Mitigating Subspace Interference via Orthogonal Adaptation also fits: Test-time compute and efficient thinking
  8. Monitorability Disposition in Large Reasoning Models also fits: Prompting and chain-of-thought reasoning
  9. PyINE: A Framework for Scalable Elicitation and Oversight via Code Execution
  10. MLLMs Fail to Refuse when Using Tools Agentically
  11. Ask, Relax, or Act? Evaluating Actionable Indeterminacy in LLM Preference Reasoning
  12. MOF-VERIFY: A Failure-Aware Agentic Harness for MOF Hypothesis Verification
  13. DeReAct: Decomposed Reasoning and Acting for Reliable AI Agents
  14. Mimir: Physics-Grounded LLM Agents for Long-Horizon Irrigation Control
  15. Learning to Ask: Information Acquisition for SLM-LLM Collaboration, under a budget also fits: Reasoning benchmarks and evaluation
  16. ReCast: Contract-Preserving Protection for Fixed-Interface Multimodal Reasoning
  17. ABDA-NL: A Natural-Language Scenario Explorer for Argument-Based Reasoning also fits: Reasoning benchmarks and evaluation
  18. Learning Meta-Skills for Agent Harness Design in Test-Time AI4AI
  19. JevSpawn: Adaptive Agentic Inference through Compositional Action Spaces
  20. OverForge: Reasoning Through Strategies and Tactics Helps Cooperative Lifelong Adaptation
  21. Thinking Outside the Box: Can Language Models Rely on External Guidance Selectively? also fits: Reinforcement learning for reasoning
  22. MADBench: Benchmarking the Security of Multi-Agent Debate
  23. Behavior-Grounded Semantic Enrichment for Financial Fraud Modeling and Reasoning
  24. ArgGYM: A Procedural, Engine-Verified Benchmark for Structured Defeasible Reasoning also fits: Mathematical reasoning
  25. EnterpriseBench: Benchmarking LLM Agents on Enterprise-Level Strategic Reasoning and Decision-Making
  26. From Judgment Quality to Downstream Utility: Rethinking LLM-as-a-Judge for Open-Ended Tasks also fits: Reinforcement learning for reasoning
  27. LatentSift: Policy-State Filtering for Token-Efficient Verification of Software Engineering Agents
  28. SCOUT: Synergizing Reasoning and Tool-Use for Computer-Use Safety
  29. Representation Alignment as a Bottleneck in LLM-Based Retrosynthesis Planning
  30. Self-Adapting Group of Experts for Multi-Agent Reasoning
  31. TMCS: Tool-Grounded Multi-Agent Reasoning for Compositional Chemical Problem Solving
  32. DeShortcut-Align: Decoupling Spurious Shortcuts for Robust Safety Alignment in Large Reasoning Models also fits: Test-time compute and efficient thinking
  33. LongPuzzleBench: Evaluating GUI Agents on Long-Horizon Visual Puzzles
  34. Codoku: Renewable Program-Reasoning Challenges for Frontier Coding Agents
  35. SGG-ReflAct: Sub-Goal Guided ReflAct with Structured Planning for Reliable Long-Horizon Reasoning also fits: Test-time compute and efficient thinking
  36. The Marathon of Scientific Reasoning: Robustness of Scientific Agents to Perturbations in Multi-Turn Interactions
  37. PowerBench: A Benchmark for Agentic Retrieval and Reasoning in Power Systems
  38. RGDT-Bench: Benchmarking LLM Reasoning for Rule-Governed Decisions and Their Justifications
  39. Improving Large Language Models for Code through Runtime Program-State Reasoning
  40. SAGE: Structured Strategic Reasoning for Efficient LLM Game Playing
  41. Beyond Solo and Consistency: Vindicating Multi-Agent Debate via Conditional Progressive Pruning
  42. Agentic Multi-Turn Reasoning: A Fairness Approach also fits: Reinforcement learning for reasoning
  43. ARSM: Auto-Regressive State Machine for Agentic Reasoning Compression also fits: Test-time compute and efficient thinking
  44. CAIRN: Dynamic Fact-Intent DAGs for Multi-Agent Exploration
  45. Beyond Prompt or Skill? Attribution-Guided Optimization of Modular LLM Programs also fits: Reinforcement learning for reasoning
  46. AuthorityLens: Rethinking LLM-Based Agent Systems Through the Lens of Authority
  47. From Trajectories to Grounded Preferences: Process Preference Synthesis via Interaction Element Graphs for Web PRMs also fits: Reinforcement learning for reasoning
  48. EmailBench: A Benchmark for Evaluating LLM Agents on Enterprise Email and Productivity Tasks
  49. Self-Play Search Distillation for Large Language Model Reasoning also fits: Mathematical reasoning
  50. When Can Agents Forget Their Reasoning? ICLR for Long-Horizon Agent Context Compression also fits: Test-time compute and efficient thinking
  51. Prefilling the Reasoning Channel: Output-Prefix Attacks on Reasoning LLMs also fits: Prompting and chain-of-thought reasoning
  52. Can LLMs Reason About Runtime Behavior? A Repository-Level Dynamic Benchmark
  53. Agent-Editing World Model: Rethinking World Modeling for LLM Agents
  54. Controlling Collectives of AI Agents in Reasoning Space with Spatial Transformers also fits: Spatial and embodied reasoning
  55. DRSR: Learning Set-Level Deletion Risk for Efficient Long-Horizon Agents also fits: Test-time compute and efficient thinking
  56. Do We Need Complex Topology Control? Distinct-Peer Random Routing Improves Cost-Efficiency in Sparse Multi-Agent Debate
  57. Stage-Supervised Latent Reasoning for Single-Shot JavaScript Deobfuscation also fits: Test-time compute and efficient thinking
  58. TelecomGPT-R1: Unified Post-Training for Reasoning Across Heterogeneous Telecom Tasks also fits: Test-time compute and efficient thinking
  59. When LLM Agents Fail to Read the Room: ReAdapt for Relational Social Reasoning
  60. GRUET: Quantifying Uncertainty of Agentic Reasoning-and-Acting Processes
  61. Reasoning Topology Matters: A Controlled Study of LLM-Based Cybersecurity Analysis
  62. Monitorable Chart Reasoning Agents via Verifiable Process Rewards also fits: Reinforcement learning for reasoning
  63. FLARE: A Full-Lifecycle Dense Supervision Paradigm for Long-Horizon Coding Agents via Generative Reward Model also fits: Reinforcement learning for reasoning
  64. On the Efficiency-Safety Dilemma in Large Reasoning Models also fits: Test-time compute and efficient thinking
  65. ISA-Bench: A Benchmark for Computational Reasoning Across Instruction Set Architectures also fits: Mathematical reasoning
  66. MIRAGE: Multi-Perspective Creative Language Model Reasoning with Reinforcement Learning Guidance also fits: Reasoning benchmarks and evaluation
  67. PetriBench: Benchmarking LLM Reasoning over Dynamic State Spaces also fits: Test-time compute and efficient thinking
  68. EviRCA: Decoupling Evidence Extraction from Reasoning for Microservice Root-Cause Analysis
  69. HEROIC: Heterogeneous Evidential Reasoning for Open-Vocabulary Identification and Cross-Robot Collaboration
  70. Evolution or Illusion? Rethinking Evaluation in LLM Evolutionary Search also fits: Reinforcement learning for reasoning
  71. Rethinking Multi-Agent Collaboration: When More Is Less
  72. AURORA: A Natural Language-Driven Agentic Framework for Understanding, Reasoning, and Orchestrating Reliable Air-Ground Co-Simulation

Foundations

Most cited by papers here

  1. OpenAI o1 System Card2024cited by 149 here
  2. LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code2024cited by 111 here
  3. Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debate2023cited by 98 here
  4. Qwen2.5-Coder Technical Report2024cited by 85 here
  5. SafeChain: Safety of Language Models with Long Chain-of-Thought Reasoning Capabilities2025cited by 75 here

Lineage

The papers this direction builds on, oldest first.

Each paper on the line cites the one before.

  1. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 648 here
  2. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 261 here
  3. 2022Large Language Models are Zero-Shot Reasonerscited by 177 here
  4. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 363 here
  5. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 263 here
  6. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 247 here
  7. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 430 here
17 more it builds on, not in this chain
  1. 2017Proximal Policy Optimization Algorithmscited by 139 here
  2. 2022Least-to-Most Prompting Enables Complex Reasoning in Large Language Modelscited by 91 here
  3. 2022Program of Thoughts Prompting: Disentangling Computation from Reasoning for Numerical Reasoning Taskscited by 96 here
  4. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 103 here
  5. 2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 184 here
  6. 2023Improving Factuality and Reasoning in Language Models through Multiagent Debatecited by 134 here
  7. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 168 here
  8. 2023Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debatecited by 98 here
  9. 2023Let's Verify Step by Stepcited by 123 here
  10. 2023Graph of Thoughts: Solving Elaborate Problems with Large Language Modelscited by 88 here
  11. 2023Code Llama: Open Foundation Models for Codecited by 99 here
  12. 2024LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Codecited by 111 here
  13. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameterscited by 92 here
  14. 2024OpenAI o1 System Cardcited by 149 here
  15. 2024DeepSeek-V3 Technical Reportcited by 126 here
  16. 2025Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learningcited by 87 here
  17. 2025Qwen3 Technical Reportcited by 273 here

The 24 most cited of 2,973.

Browse

30 of 2,143, ranked 6–35

  1. 6Competition-Level Code Generation with AlphaCode202271
  2. 7Deliberative Alignment: Reasoning Enables Safer Language Models202466
  3. 8Reasoning with Language Model is Planning with World Model202366
  4. 9LegalBench: A Collaboratively Built Benchmark for Measuring Legal Reasoning in Large Language Models202361
  5. 10Teaching Large Language Models to Self-Debug202353
  6. 11The Hidden Risks of Large Reasoning Models: A Safety Assessment of R1202552
  7. 12ReTool: Reinforcement Learning for Strategic Tool Use in LLMs202550
  8. 13Language Agent Tree Search Unifies Reasoning Acting and Planning in Language Models202349
  9. 14H-CoT: Hijacking the Chain-of-Thought Safety Reasoning Mechanism to Jailbreak Large Reasoning Models, Including OpenAI o1/o3, DeepSeek-R1, and Gemini 2.0 Flash Thinking202542
  10. 15STAR-1: Safer Alignment of Reasoning LLMs with 1K Data202539
All 30
  1. 16Mixture-of-Agents Enhances Large Language Model Capabilities202437
  2. 17Safety Tax: Safety Alignment Makes Your Large Reasoning Models Less Reasonable202536
  3. 18ReConcile: Round-Table Conference Improves Reasoning via Consensus among Diverse LLMs202336
  4. 19PlanBench: An Extensible Benchmark for Evaluating Large Language Models on Planning and Reasoning about Change202234
  5. 20How Johnny Can Persuade LLMs to Jailbreak Them: Rethinking Persuasion to Challenge AI Safety by Humanizing LLMs202431
  6. 21GuardReasoner: Towards Reasoning-based LLM Safeguards202530
  7. 22Fin-R1: A Large Language Model for Financial Reasoning through Reinforcement Learning202530
  8. 23STAIR: Improving Safety Alignment with Introspective Reasoning202529
  9. 24RealSafe-R1: Safety-Aligned DeepSeek-R1 without Compromising Reasoning Capability202528
  10. 25Reasoning Runtime Behavior of a Program with LLM: How Far Are We?202428
  11. 26OverThink: Slowdown Attacks on Reasoning LLMs202527
  12. 27AFlow: Automating Agentic Workflow Generation202427
  13. 28Structured Chain-of-Thought Prompting for Code Generation202327
  14. 29ConvFinQA: Exploring the Chain of Numerical Reasoning in Conversational Finance Question Answering202227
  15. 30Safety in Large Reasoning Models: A Survey202526
  16. 31BadChain: Backdoor Chain-of-Thought Prompting for Large Language Models202425
  17. 32LLM as a Mastermind: A Survey of Strategic Reasoning with Large Language Models202425
  18. 33Reasoning-to-Defend: Safety-Aware Reasoning Can Defend Large Language Models from Jailbreaking202524
  19. 34GTBench: Uncovering the Strategic Reasoning Limitations of LLMs via Game-Theoretic Evaluations202424
  20. 35Chain of Code: Reasoning with a Language Model-Augmented Code Emulator202324