mlsanityData to 7 Oct 2026

BrowseConf: Confidence-Guided Test-Time Scaling for Web Agents

arXiv 2510.23458 ↗First posted 27 Oct 2025cs.CL cs.AIcited by 5

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

also in 1 more
Abstract (arXiv)

Confidence in LLMs is a useful indicator of model uncertainty and answer reliability. Existing work mainly focused on single-turn scenarios, while research on confidence in complex multi-turn interactions is limited. In this paper, we investigate whether LLM-based search agents have the ability to communicate their own confidence through verbalized confidence scores after long sequences of actions, a significantly more challenging task compared to outputting confidence in a single interaction. Experimenting on open-source agentic models, we first find that models exhibit much higher task accuracy at high confidence while having near-zero accuracy when confidence is low. Based on this observation, we propose Test-Time Scaling (TTS) methods that use confidence scores to determine answer quality, encourage the model to try again until reaching a satisfactory confidence level. Results show that our proposed methods significantly reduce token consumption while demonstrating competitive performance compared to baseline fixed budget TTS methods.

Key facts. Cited by 5 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 39 of its 48 references that are in the map. First posted on arXiv 27 Oct 2025.

Builds on

39 of 48 references.

Closest ancestors · 2

Same direction, newest first

  1. 2025WebSailor: Navigating Super-human Reasoning for Web Agent3 Jul
  2. 2025ToLeaP: Rethinking Development of Tool Learning with Large Language Models17 May

Other references · 37

  1. 2023GPT-4 Technical Reportcited by 18,132
  2. 2015Distilling the Knowledge in a Neural Networkcited by 10,076
  3. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  4. 2017On Calibration of Modern Neural Networkscited by 4,860
  5. 2022Language Models (Mostly) Know What They Knowcited by 1,783
  6. 2024τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domainscited by 1,280
  7. 2025gpt-oss-120b & gpt-oss-20b Model Cardcited by 1,258
  8. 2023Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedbackcited by 788
  9. 2024DeepSeek-V3 Technical Reportcited by 724
  10. 2025BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agentscited by 616
  11. 2025τ2-Bench: Evaluating Conversational Agents in a Dual-Control Environmentcited by 558
  12. 2025Humanity's Last Examcited by 467
25 more
  1. 2019Measuring Calibration in Deep Learningcited by 443
  2. 2024Semantic Entropy Probes: Robust and Cheap Hallucination Detection in LLMscited by 246
  3. 2025A Survey on Test-Time Scaling in Large Language Models: What, How, Where, and How Well?cited by 215
  4. 2025Why Language Models Hallucinatecited by 205
  5. 2025WebWalker: Benchmarking LLMs in Web Traversalcited by 171
  6. 2025BrowseComp-ZH: Benchmarking Web Browsing Ability of Large Language Models in Chinesecited by 139
  7. 2024Taming Overconfidence in LLMs: Reward Calibration in RLHFcited by 105
  8. 2024Large Language Models Must Be Taught to Know What They Don't Knowcited by 99
  9. 2025Maximizing Confidence Alone Improves Reasoningcited by 97
  10. 2024On Verbalized Confidence Scores for LLMscited by 83
  11. 2025Efficient Test-Time Scaling via Self-Calibrationcited by 74
  12. 2025Beyond Single-Turn: A Survey on Multi-Turn Interactions with Large Language Modelscited by 58
  13. 2024Calibrating the Confidence of Large Language Models by Eliciting Fidelitycited by 47
  14. 2025Think Twice: Enhancing LLM Reasoning by Scaling Multi-round Test-time Thinkingcited by 46
  15. 2024When to Trust LLMs: Aligning Confidence with Response Qualitycited by 42
  16. 2025Mind the Confidence Gap: Overconfidence, Calibration, and Distractor Effects in Large Language Modelscited by 39
  17. 2025Rethinking Fine-Tuning when Scaling Test-Time Compute: Limiting Confidence Improves Mathematical Reasoningcited by 28
  18. 2025TokUR: Token-Level Uncertainty Estimation for Large Language Model Reasoningcited by 24
  19. 2025SteerConf: Steering LLMs for Confidence Elicitationcited by 20
  20. 2024SAUP: Situation Awareness Uncertainty Propagation on LLM Agentcited by 14
  21. 2025Generative AI Act II: Test Time Scaling Drives Cognition Engineeringcited by 14
  22. 2024Graph-based Confidence Calibration for Large Language Modelscited by 10
  23. 2024Calibrating Verbalized Probabilities for Large Language Modelscited by 9
  24. 2023Methods to Estimate Large Language Model Confidencecited by 4
  25. 2025GrACE: A Generative Approach to Better Confidence Elicitation and Efficient Test-Time Scaling in Large Language Modelscited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 5.

5 papers.

  1. 2026Uncertainty Decomposition for Clarification Seeking in LLM Agentscited by 3
  2. 2026Last Step Matters: Early Uncertainty Cannot Predict Failure in Long-Horizon Agentscited by 2
  3. 2026Mobile-Aptus: Confidence-Driven Proactive and Robust Interaction in MLLM-based Mobile-Using Agentscited by 0
  4. 2026Beyond Confidence: Test-Time Scaling for Multi-Turn Search Agents via Retrieval Groundingcited by 0
  5. 2026What Can Component-Replacement Evidence Establish? A Critical Scoping Review of Local Decisions in LLM Agentscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Early-Token Confidence Predicts Reasoning Quality in Multi-Agent LLM Debate
  2. 2026Benchmark Test-Time Scaling of General LLM Agents
  3. 2026Agentic Test-Time Scaling for WebAgents
  4. 2026Scaling with Confidence: Calibrating Confidence of LLMs for Adaptive Test Time Scaling
  5. 2026Uncertainty Quantification for LLM Agents: A Taxonomy, an Evaluation Protocol, and an Empirical Study
  6. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters
  7. 2026Evaluating Confidence-Gated Retrieval with Matched Trajectory Replay
  8. 2026Confidence Composition for Multiagent Language Model Systems