mlsanityData to 14 Oct 2026

ConfidenceBench: Evaluating Confidence Calibration in Large Language Models

arXiv 2607.20526 ↗First posted 10 Jul 2026cs.AI cs.LG stat.MLcited by 2

Not on the map. Why?

Abstract (arXiv)

Large language models (LLMs) are increasingly deployed in settings where fluent but incorrect answers can be costly. In these settings, accuracy alone is insufficient: models must also know when they are likely to be wrong. We present ConfidenceBench, a calibration benchmark that evaluates verbalized confidence estimates in 15 frontier LLMs using the Brier score, a proper scoring rule that incentivises truthful probability reporting. Confidence is elicited via prompting, requiring no access to model logits and making the framework applicable to both closed-source and open-source systems. The benchmark comprises 200 private multiple-choice questions across four categories: spatial reasoning, high-precision mathematics, word lookup, and unknowable questions. Across three independent runs, Claude Opus 4.6 and Gemini 3.1 Pro Preview achieve the lowest Brier scores, both reported as 0.103. Both substantially outperform the calibrated-random baseline of 0.1875, while Gemini 3.1 Flash-Lite scores 0.367, indicating severe miscalibration. Accuracy and calibration diverge substantially across model families: the most accurate model is not the best-calibrated, and several models perform worse than the calibrated-random Brier baseline despite reasonable accuracy. These results show that verbalized confidence calibration is a distinct and practically important axis of LLM reliability, complementary to standard accuracy-based evaluation.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 22 of its 27 references that are in the map. First posted on arXiv 10 Jul 2026.

Builds on

22 of 27 references.

References · 22

  1. 2022Training language models to follow instructions with human feedbackcited by 15,971
  2. 2020Measuring Massive Multitask Language Understandingcited by 7,963
  3. 2017On Calibration of Modern Neural Networkscited by 4,960
  4. 2017Deep reinforcement learning from human preferencescited by 3,912
  5. 2022Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language modelscited by 1,998
  6. 2022Language Models (Mostly) Know What They Knowcited by 1,856
  7. 2023Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMscited by 974
  8. 2023Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedbackcited by 812
  9. 2022Teaching Models to Express Their Uncertainty in Wordscited by 751
  10. 2019Verified Uncertainty Calibrationcited by 309
  11. 2023Do Large Language Models Know What They Don't Know?cited by 243
  12. 2025AbstentionBench: Reasoning LLMs Fail on Unanswerable Questionscited by 145
10 more
  1. 2023LM-Polygraph: Uncertainty Estimation for Language Modelscited by 131
  2. 2024Benchmarking LLMs via Uncertainty Quantificationcited by 126
  3. 2024Large Language Models Must Be Taught to Know What They Don't Knowcited by 100
  4. 2025Uncertainty Quantification and Confidence Calibration in Large Language Models: A Surveycited by 95
  5. 2024On Verbalized Confidence Scores for LLMscited by 83
  6. 2023Smooth ECE: Principled Reliability Diagrams via Kernel Smoothingcited by 63
  7. 2024MAQA: Evaluating Uncertainty Quantification in LLMs Regarding Data Uncertaintycited by 27
  8. 2024UBench: Benchmarking Uncertainty in Large Language Models with Multiple Choice Questionscited by 7
  9. 2025Torch-Uncertainty: A Deep Learning Framework for Uncertainty Quantificationcited by 4
  10. 2026Confidence-Calibrated Small-Large Language Model Collaboration for Cost-Efficient Reasoningcited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026Confident at the moment of action: belief miscalibration in LLM play under hidden informationcited by 1
  2. 2026Replication Without Persistence in Hosted LLMs: Measurement Sensitivity in Action-Time Belief Evaluationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024Calibrating the Confidence of Large Language Models by Eliciting Fidelity
  2. 2026CALIBER: Calibrating Confidence Before and After Reasoning in Language Models
  3. 2025SteerConf: Steering LLMs for Confidence Elicitation
  4. 2025Mind the Confidence Gap: Overconfidence, Calibration, and Distractor Effects in Large Language Models
  5. 2026Rethinking Uncertainty Evaluation in Large Language Models
  6. 2024Confidence in the Reasoning of Large Language Models
  7. 2024Confidence Under the Hood: An Investigation into the Confidence-Probability Alignment in Large Language Models
  8. 2024Calibrating Long-form Generations from Large Language Models