mlsanityData to 7 Oct 2026

Why Reasoning Fails to Plan: A Planning-Centric Analysis of Long-Horizon Decision Making in LLM Agents

arXiv 2601.22311 ↗First posted 29 Jan 2026cs.AI cs.CL cs.LGcited by 22

On the map: LLM reasoning → Reasoning for agents, planning and code

also in 1 more
Abstract (arXiv)

Large language model (LLM)-based agents exhibit strong step-by-step reasoning capabilities over short horizons, yet often fail to sustain coherent behavior over long planning horizons. We argue that this failure reflects a fundamental mismatch: step-wise reasoning induces a form of step-wise greedy policy that is adequate for short horizons but fails in long-horizon planning, where early actions must account for delayed consequences. From this planning-centric perspective, we study LLM-based agents in deterministic, fully structured environments with explicit state transitions and evaluation signals. Our analysis reveals a core failure mode of reasoning-based policies: locally optimal choices induced by step-wise scoring lead to early myopic commitments that are systematically amplified over time and difficult to recover from. We introduce FLARE (Future-aware Lookahead with Reward Estimation) as a minimal instantiation of future-aware planning to enforce explicit lookahead, value propagation, and limited commitment in a single model, allowing downstream outcomes to influence early decisions. Across multiple benchmarks, agent frameworks, and LLM backbones, FLARE consistently improves task performance and planning-level behavior, frequently allowing LLaMA-8B with FLARE to outperform GPT-4o with standard step-by-step reasoning. These results establish a clear distinction between reasoning and planning.

Key facts. Cited by 22 papers in the map (data through 7 Oct 2026). Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning. Builds on 46 of its 123 references that are in the map. First posted on arXiv 29 Jan 2026.

Builds on

46 of 123 references.

Closest ancestors · 3

Same direction, newest first

  1. 2024Non-myopic Generation of Language Models for Reasoning and Planning22 Oct
  2. 2024GuardAgent: Safeguard LLM Agents by a Guard Agent via Knowledge-Enabled Reasoning13 Jun
  3. 2023Language Models, Agent Models, and World Models: The LAW for Machine Reasoning and Planning8 Dec

Other references · 43

  1. 2024The Llama 3 Herd of Modelscited by 14,192
  2. 2022Do As I Can, Not As I Say: Grounding Language in Robotic Affordancescited by 2,302
  3. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameterscited by 1,928
  4. 2023ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIscited by 1,880
  5. 2025Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learningcited by 1,371
  6. 2023AgentBench: Evaluating LLMs as Agentscited by 1,211
  7. 2022Inner Monologue: Embodied Reasoning through Planning with Language Modelscited by 1,057
  8. 2023Large Language Models Cannot Self-Correct Reasoning Yetcited by 965
  9. 2023A Multitask, Multilingual, Multimodal Evaluation of ChatGPT on Reasoning, Hallucination, and Interactivitycited by 888
  10. 2025Why Do Multi-Agent LLM Systems Fail?cited by 559
  11. 2025LLMs Get Lost In Multi-Turn Conversationcited by 400
  12. 2023The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"cited by 399
31 more
  1. 2023Trustworthy LLMs: a Survey and Guideline for Evaluating Large Language Models' Alignmentcited by 394
  2. 2025From System 1 to System 2: A Survey of Reasoning Large Language Modelscited by 319
  3. 2025Towards Large Reasoning Models: A Survey of Reinforced Reasoning with Large Language Modelscited by 291
  4. 2025Reinforcement Learning for Reasoning in Large Language Models with One Training Examplecited by 288
  5. 2024CodeAgent: Enhancing Code Generation with Tool-Integrated Agent Systems for Real-World Repo-level Coding Challengescited by 278
  6. 2023Automatically Correcting Large Language Models: Surveying the landscape of diverse self-correction strategiescited by 236
  7. 2023Large Language Model Guided Tree-of-Thoughtcited by 218
  8. 2025Plan-and-Act: Improving Planning of Agents for Long-Horizon Taskscited by 190
  9. 2025Beyond Ten Turns: Unlocking Long-Horizon Agentic Search with Large-Scale Asynchronous RLcited by 144
  10. 2024StableToolBench: Towards Stable Large-Scale Benchmarking on Tool Learning of Large Language Modelscited by 129
  11. 2025Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systemscited by 125
  12. 2025AgentGym-RL: Training LLM Agents for Long-Horizon Decision Making through Multi-Turn Reinforcement Learningcited by 68
  13. 2024Think-on-Graph 2.0: Deep and Faithful Large Language Model Reasoning with Knowledge-guided Retrieval Augmented Generationcited by 63
  14. 2023Beyond Chain-of-Thought, Effective Graph-of-Thought Reasoning in Language Modelscited by 51
  15. 2024FlowBench: Revisiting and Benchmarking Workflow-Guided Planning for LLM-based Agentscited by 39
  16. 2025Monte Carlo Tree Diffusion for System 2 Planningcited by 33
  17. 2025AgentRouter: A Knowledge-Graph-Guided LLM Router for Collaborative Multi-Agent Question Answeringcited by 29
  18. 2025AutoData: A Multi-Agent System for Open Web Data Collectioncited by 25
  19. 2024Beyond Performance: Quantifying and Mitigating Label Bias in LLMscited by 24
  20. 2025MLA-Trust: Benchmarking Trustworthiness of Multimodal LLM Agents in GUI Environmentscited by 23
  21. 2025Self-Correction Bench: Uncovering and Addressing the Self-Correction Blind Spot in Large Language Modelscited by 19
  22. 2025World Modelling Improves Language Model Agentscited by 12
  23. 2024Tree-of-Traversals: A Zero-Shot Reasoning Algorithm for Augmenting Black-box Language Models with Knowledge Graphscited by 11
  24. 2025DeepSearch: Overcome the Bottleneck of Reinforcement Learning with Verifiable Rewards via Monte Carlo Tree Searchcited by 11
  25. 2025WorldLLM: Improving LLMs' world modeling using curiosity-driven theory-makingcited by 10
  26. 2025DAMR: Efficient and Adaptive Context-Aware Knowledge Graph Question Answering with LLM-Guided MCTScited by 6
  27. 2025Semantic Refinement with LLMs for Graph Representationscited by 6
  28. 2026LongDA: Benchmarking LLM Agents for Long-Document Data Analysiscited by 6
  29. 2025Controllable Graph Generation with Diffusion Models via Inference-Time Tree Search Guidancecited by 5
  30. 2025Enhancing Large Language Models with Reward-guided Tree Search for Knowledge Graph Question and Answeringcited by 3
  31. 2025ProgRAG: Hallucination-Resistant Progressive Retrieval and Reasoning over Knowledge Graphscited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 22.

22 papers.

Sort:

By area:

  1. 2026Why Semantic Entropy Fails: Geometry-Aware and Calibrated Uncertainty for Policy Optimizationcited by 5
  2. 2026W2T: LoRA Weights Already Know What They Can Docited by 4
  3. 2026TACT: Mitigating Overthinking and Overacting in Coding Agents via Activation Steeringcited by 4
  4. 2026Same Signal, Opposite Meaning: Direction-Informed Adaptive Learning for LLM Agentscited by 4
  5. 2026PreScam: A Benchmark for Predicting Scam Progression from Early Conversationscited by 2
  6. 2026MDForge: Agentic Molecular Dynamics Pipeline Design under Sparse Simulator Feedbackcited by 2
  7. 2026Skill Reuse as Compression in Agentic RLcited by 1
  8. 2026ProPlay: Procedural World Models for Self-Evolving LLM Agentscited by 1
  9. 2026Towards Reliable and Robust LLM Planning: Symbolic Feedback-Driven Iterative Self-Refinement Frameworkcited by 1
  10. 2026Ask the World Before Acting: Environment Probing for Calibrated Agent World Modelscited by 1
  11. 2026AgenticSTS: A Bounded-Memory Testbed for Long-Horizon LLM Agentscited by 1
  12. 2026The Collaboration Tax: How Much LLM Multi-Agent Systems Pay to Coordinatecited by 1
10 more
  1. 2026CAFE: Self-Improving Search Agents Need Co-Evolving Feedbackcited by 1
  2. 2026Polished but Unresolved: Identifying Late-Stage Pressure States in Long-Horizon Tool-Use Agentscited by 1
  3. 2026Calibration is the Bottleneck: An Action-Class Diagnostic of Multi-Turn Tool-Callingcited by 1
  4. 2026MineEvolve: Self-Evolution with Accumulated Knowledge for Long-Horizon Embodied Minecraft Agentscited by 0
  5. 2026Emergent Strategic Reasoning Risks in AI: A Taxonomy-Driven Evaluation Frameworkcited by 0
  6. 2026Learning Stateful Predictive Knowledge From Experiencecited by 0
  7. 2026Verifiable Hidden Dynamics Play: Generating Agentic RL Environments from Solved Mechanismscited by 0
  8. 2026Verification as an Architectural Layer for LLM Agents: A V-Model Design, and a Pilot Study of Its Deterministic Corecited by 0
  9. 2026HyperMCTS: Hypergraph-Augmented MCTS for Long-Horizon LLM Agentscited by 0
  10. 2026Marathoner: Ultra-Long-Horizon Autonomous Intelligencecited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Robotouille: An Asynchronous Planning Benchmark for LLM Agents
  2. 2025Learning When to Plan: Efficiently Allocating Test-Time Compute for LLM Agents
  3. 2026Extracting Search Trees from LLM Reasoning Traces Reveals Myopic Planning
  4. 2026Efficient Agentic Reasoning Through Self-Regulated Simulative Planning
  5. 2026SokoBench: Evaluating Long-Horizon Planning and Reasoning in Large Language Models
  6. 2024Chasing Progress, Not Perfection: Revisiting Strategies for End-to-End LLM Plan Generation
  7. 2026Planner Matters! An Efficient and Unbalanced Multi-agent Collaboration Framework for Long-horizon Planning
  8. 2024Can We Rely on LLM Agents to Draft Long-Horizon Plans? Let's Take TravelPlanner as an Example