Revisiting Uncertainty Estimation and Calibration of Large Language Models

TL;DR

This comprehensive study evaluates uncertainty estimation methods across 80 large language models, revealing that linguistic verbal uncertainty (LVU) offers superior calibration and interpretability, and emphasizing the complex factors influencing uncertainty reliability.

Contribution

The paper provides the most extensive evaluation of uncertainty estimation in LLMs, comparing three black-box methods across diverse models and tasks, and highlights LVU as a practical, effective approach.

Findings

LVU outperforms TPU and NVU in calibration and interpretability.

High accuracy does not guarantee reliable uncertainty estimates.

Model scale, reasoning ability, and quantization affect uncertainty performance.

Abstract

As large language models (LLMs) are increasingly deployed in high-stakes applications, robust uncertainty estimation is essential for ensuring the safe and trustworthy deployment of LLMs. We present the most comprehensive study to date of uncertainty estimation in LLMs, evaluating 80 models spanning open- and closed-source families, dense and Mixture-of-Experts (MoE) architectures, reasoning and non-reasoning modes, quantization variants and parameter scales from 0.6B to 671B. Focusing on three representative black-box single-pass methods, including token probability-based uncertainty (TPU), numerical verbal uncertainty (NVU), and linguistic verbal uncertainty (LVU), we systematically evaluate uncertainty calibration and selective classification using the challenging MMLU-Pro benchmark, which covers both reasoning-intensive and knowledge-based tasks. Our results show that LVU consistently outperforms TPU and NVU, offering stronger calibration and discrimination while being more interpretable. We also find that high accuracy does not imply reliable uncertainty, and that model scale, post-training, reasoning ability and quantization all influence estimation performance. Notably, LLMs exhibit better uncertainty estimates on reasoning tasks than on knowledge-heavy ones, and good calibration does not necessarily translate to effective error ranking. These findings highlight the need for multi-perspective evaluation and position LVU as a practical tool for improving the reliability of LLMs in real-world settings.