mlsanityData to 14 Oct 2026

LEAD: Breaking the No-Recovery Bottleneck in Long-Horizon Reasoning

arXiv 2603.06870 ↗First posted 6 Mar 2026cs.AIcited by 1

On the map: LLM reasoning → Test-time compute and efficient thinking

Abstract (arXiv)

Long-horizon execution in Large Language Models (LLMs) remains unstable even when high-level strategies are provided. Evaluating on controlled algorithmic puzzles, we demonstrate that while decomposition is essential for stability, extreme decomposition creates a "no-recovery bottleneck". We show that this bottleneck becomes critical due to highly non-uniform error distribution, where consistent errors on a few "hard" steps become irreversible. To address this, we propose Lookahead-Enhanced Atomic Decomposition (LEAD). By incorporating short-horizon future validation and aggregating overlapping rollouts, LEAD provides enough isolation to maintain stability while retaining enough local context to correct errors. This enables the o4-mini model to solve Checkers Jumping up to complexity n=13, whereas extreme decomposition fails beyond n=11.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 18 of its 40 references that are in the map. First posted on arXiv 6 Mar 2026.

Builds on

18 of 40 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025The Markovian Thinker: Architecture-Agnostic Linear Scaling of Reasoning8 Oct
  2. 2025The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity7 Jun
  3. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Models17 May

Other references · 15

  1. 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,455
  2. 2022Least-to-Most Prompting Enables Complex Reasoning in Large Language Modelscited by 1,407
  3. 2022Measuring and Narrowing the Compositionality Gap in Language Modelscited by 1,075
  4. 2023Plan-and-Solve Prompting: Improving Zero-Shot Chain-of-Thought Reasoning by Large Language Modelscited by 560
  5. 2019Lookahead Optimizer: k steps forward, 1 step backcited by 336
  6. 2023What Algorithms can Transformers Learn? A Study in Length Generalizationcited by 199
  7. 2025Artificial Hivemind: The Open-Ended Homogeneity of Language Models (and Beyond)cited by 141
  8. 2025Context Length Alone Hurts LLM Performance Despite Perfect Retrievalcited by 127
  9. 2025The Illusion of Diminishing Returns: Measuring Long Horizon Execution in LLMscited by 64
  10. 2024MegaAgent: A Large-Scale Autonomous LLM-based Multi-Agent System Without Predefined SOPscited by 47
  11. 2025Context as a Tool: Context Management for Long-Horizon SWE-Agentscited by 43
  12. 2024Not All LLM Reasoners Are Created Equalcited by 30
3 more
  1. 2024ALR2: A Retrieve-then-Reason Framework for Long-context Question Answeringcited by 26
  2. 2025Solving a Million-Step LLM Task with Zero Errorscited by 19
  3. 2025Positional Biases Shift as Inputs Approach Context Window Limitscited by 9

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Towards a Systems Foundation for Agentic Skills: Architecture, Lifecycle, and Securitycited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Empirical Evidence of Complexity-Induced Limits in Large Language Models on Finite Discrete State-Space Problems with Explicit Validity Constraints
  2. 2025seqBench: A Tunable Benchmark to Quantify Sequential Reasoning Limits of LLMs
  3. 2025Beyond Exponential Decay: Rethinking Error Accumulation in Large Language Models
  4. 2026A Decomposition Perspective to Long-context Reasoning for LLMs
  5. 2026SokoBench: Evaluating Long-Horizon Planning and Reasoning in Large Language Models
  6. 2026Intrinsic Stability Limits of Autoregressive Reasoning: Structural Consequences for Long-Horizon Execution
  7. 2025From Implicit Exploration to Structured Reasoning: Leveraging Guideline and Refinement for LLMs
  8. 2026Wrong Prediction, Right Answer: Recovering Evidence from Collapsed LLM Sequence Scores