mlsanityData to 7 Oct 2026

Measuring Weak-to-Strong Legibility of Reasoning Models

arXiv 2603.20508 ↗First posted 20 Mar 2026cs.MA cs.AI cs.CLcited by 5

On the map: LLM reasoning → Prompting and chain-of-thought reasoning

Abstract (arXiv)

Reasoning language models (RLMs) and the intermediate chains of thought they emit play an increasingly central role in multi-agent setups such as inter-model monitoring or distillation into smaller models. When agents at different capability tiers must cooperate, strong models need to produce traces digestible by weaker ones. We refer to this goal as "weak-to-strong legibility". Trustworthiness of large models depends in part on this legibility property. For safety oversight in particular, adoption of weak monitors may become a standard for reliability scaffolds on a healthy budget. Legibility requires that the shape of these decision-making traces takes some form accessible to weaker monitors. Existing efficiency-based metrics for legibility fail to capture "thoroughness", instead focusing on conciseness.

Key facts. Cited by 5 papers in the map (data through 7 Oct 2026). Placed in Prompting and chain-of-thought reasoning, one of 17 directions in LLM reasoning. Builds on 39 of its 60 references that are in the map. First posted on arXiv 20 Mar 2026.

Builds on

39 of 60 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025A Pragmatic Way to Measure Chain-of-Thought Monitorability28 Oct
  2. 2025Do Cognitively Interpretable Reasoning Traces Improve LLM Performance?21 Aug
  3. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Models28 Jan

Other references · 36

  1. 2024The Llama 3 Herd of Modelscited by 14,192
  2. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  3. 2025Qwen3 Technical Reportcited by 8,195
  4. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,146
  5. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
  6. 2023Let's Verify Step by Stepcited by 3,598
  7. 2024Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phonecited by 1,883
  8. 2025Gemma 3 Technical Reportcited by 1,686
  9. 2025gpt-oss-120b & gpt-oss-20b Model Cardcited by 1,258
  10. 2024Tulu 3: Pushing Frontiers in Open Language Model Post-Trainingcited by 988
  11. 2025Kimi K2: Open Agentic Intelligencecited by 523
  12. 2023Weak-to-Strong Generalization: Eliciting Strong Capabilities With Weak Supervisioncited by 445
24 more
  1. 2024Training Language Models to Self-Correct via Reinforcement Learningcited by 405
  2. 2024Skywork-Reward: Bag of Tricks for Reward Modeling in LLMscited by 322
  3. 2025TokenSkip: Controllable Chain-of-Thought Compression in LLMscited by 246
  4. 2025Llama-Nemotron: Efficient Reasoning Modelscited by 151
  5. 2025Multi-Agent Risks from Advanced AIcited by 150
  6. 2024Concise Thoughts: Impact of Output Length on LLM Reasoning and Costcited by 108
  7. 2025Olmo 3cited by 95
  8. 2024Prover-Verifier Games improve legibility of LLM outputscited by 79
  9. 2025Evaluating Step-by-step Reasoning Traces: A Surveycited by 62
  10. 2025Magistralcited by 57
  11. 2025BREAD: Branched Rollouts from Expert Anchors Bridge SFT & RL for Reasoningcited by 57
  12. 2025Retro-Search: Exploring Untaken Paths for Deeper and Efficient Reasoningcited by 32
  13. 2025Metacognitive Reuse: Turning Recurring LLM Reasoning Into Concise Behaviorscited by 28
  14. 2025To Backtrack or Not to Backtrack: When Sequential Search Limits Model Reasoningcited by 24
  15. 2025An Empirical Study of LLM Reasoning Ability Under Strict Output Length Constraintcited by 16
  16. 2025Not All Thoughts are Generated Equal: Efficient LLM Reasoning via Multi-Turn Reinforcement Learningcited by 16
  17. 2025Reconsidering Overthinking: Penalizing Internal and External Redundancy in CoT Reasoningcited by 13
  18. 2025Retrieval-of-Thought: Efficient Reasoning via Reusing Thoughtscited by 12
  19. 2026How does information access affect LLM monitors' ability to detect sabotage?cited by 12
  20. 2025Local Coherence or Global Validity? Investigating RLVR Traces in Math Domainscited by 5
  21. 2026Do explanations generalize across large reasoning models?cited by 5
  22. 2026Basic Legibility Protocols Improve Trusted Monitoringcited by 3
  23. 2026Beyond Correctness: Learning Robust Reasoning via Transfercited by 2
  24. 2025Alice: Proactive Learning with Teacher's Demonstrations for Weak-to-Strong Generalizationcited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 5.

5 papers.

  1. 2026How Much Does a Reasoning Summary Reveal? An Observability Ladder for Large Language Modelscited by 1
  2. 2026CLORE: Content-Level Optimization for Reasoning Efficiencycited by 0
  3. 2026Reasoning that Travels: Dissecting How Chain-of-Thought Transfers Across Modelscited by 0
  4. 2026Tandem Reinforcement Learning with Verifiable Rewardscited by 0
  5. 2026On Language Drift during RLVR Post-Trainingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Reasoning Models Sometimes Output Illegible Chains of Thought
  2. 2026When Small Models Are Right for Wrong Reasons: Process Verification for Trustworthy Agents
  3. 2026Risky Business: Measuring The Faithfulness-Safety Tension
  4. 2026Legibility is Not Interpretability: Comparing Judged and Actual Importance in Chain-Of-Thought Reasoning
  5. 2026Project Ariadne: A Structural Causal Framework for Auditing Faithfulness in LLM Agents
  6. 2025Investigating CoT Monitorability in Large Reasoning Models
  7. 2026When to Trust the Cheap Check: Weak and Strong Verification for Reasoning
  8. 2025Monitoring Monitorability