mlsanityData to 14 Oct 2026

Certainty-Guided Reasoning in Large Language Models: A Dynamic Thinking Budget Approach

arXiv 2509.07820 ↗First posted 9 Sep 2025cs.AI cs.CLcited by 3

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

Abstract (arXiv)

Large reasoning language models are typically run with fixed inference budgets, which can waste computation or terminate reasoning prematurely. We introduce Certainty-Guided Reasoning (CGR), a model-agnostic adaptive inference procedure that periodically probes whether the current reasoning supports a confident final answer and terminates early once a target certainty threshold is reached, otherwise continuing until the end-of-thinking token or the budget limit. Certainty is estimated from the model's predicted probabilities over the answer tokens, yielding a lightweight stopping criterion. On AIME2025, CGR preserves baseline accuracy while reducing token usage, providing a tunable certainty-efficiency trade-off that can eliminate millions of tokens in aggregate. Across 64 random seeds, CGR exhibits consistent behavior. We also introduce a Grade metric that penalizes incorrect answers and permits abstention, capturing risk-sensitive performance. Results show that CGR improves Grade by abstaining when certainty remains low.

Key facts. Cited by 3 papers 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 13 of its 15 references that are in the map. First posted on arXiv 9 Sep 2025.

Builds on

13 of 15 references.

Closest ancestors · 5

Same direction, newest first

  1. 2025Inverse Scaling in Test-Time Compute19 Jul
  2. 2025Steering LLM Thinking with Budget Guidance16 Jun
  3. 2025Dynamic Early Exit in Reasoning Models22 Apr
  4. 2025s1: Simple test-time scaling31 Jan
  5. 2024Escape Sky-high Cost: Early-stopping Self-Consistency for Multi-step Reasoning19 Jan

Other references · 8

  1. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  2. 2022Generative Adversarial Networkscited by 5,627
  3. 2023Gorilla: Large Language Model Connected with Massive APIscited by 1,332
  4. 2020DeeBERT: Dynamic Early Exiting for Accelerating BERT Inferencecited by 332
  5. 2024LayerSkip: Enabling Early Exit Inference and Self-Speculative Decodingcited by 307
  6. 2025Phi-4-reasoning Technical Reportcited by 151
  7. 2024A Survey on Uncertainty Quantification of Large Language Models: Taxonomy, Open Research Challenges, and Future Directionscited by 121
  8. 2025Maximizing Confidence Alone Improves Reasoningcited by 100

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026Limited Reasoning Space: The cage of long-horizon reasoning in LLMscited by 3
  2. 2026AdaTurn: Budget-Aware Test-Time Scaling for Active Visual Perception Agentscited by 0
  3. 2026Certainty Is Not Just Correctness: Rethinking Token-Level Certainty in LLM Reasoningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Improving reasoning at inference time via uncertainty minimisation
  2. 2025Prolonged Reasoning Is Not All You Need: Certainty-Based Adaptive Routing for Efficient LLM/MLLM Reasoning
  3. 2025Efficient Reasoning for Large Reasoning Language Models via Certainty-Guided Reflection Suppression
  4. 2026CAT: Confidence-Adaptive Thinking for Efficient Reasoning of Large Reasoning Models
  5. 2022Selection-Inference: Exploiting Large Language Models for Interpretable Logical Reasoning
  6. 2026Enhancing reasoning accuracy in large language models during inference time
  7. 2026Reasoning While Asking: Transforming Reasoning Large Language Models from Passive Solvers to Proactive Inquirers
  8. 2025Advancing Reasoning in Large Language Models: Promising Methods and Approaches