# VL-Calibration: Decoupled Confidence Calibration for Large Vision-Language Models Reasoning

Source: [VL-Calibration: Decoupled Confidence Calibration for Large Vision-Language Models Reasoning](<https://www.sota2.com/research/paper/69dc78a77d0172140ec78c5a>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-14T05:00:11.147Z

Authors: Wenyi Xiao, Xinchi Xu, Leilei Gan

Published: 2026-04-10T17:47:19.000Z

## Abstract

Large Vision Language Models (LVLMs) achieve strong multimodal reasoning but frequently exhibit hallucinations and incorrect responses with high certainty, which hinders their usage in high-stakes domains. Existing verbalized confidence calibration methods, largely developed for text-only LLMs, typically optimize a single holistic confidence score using binary answer-level correctness. This design is mismatched to LVLMs: an incorrect prediction may arise from perceptual failures or from reasoning errors given correct perception, and a single confidence conflates these sources while visual uncertainty is often dominated by language priors. To address these issues, we propose VL-Calibration, a reinforcement learning framework that explicitly decouples confidence into visual and reasoning confidence. To supervise visual confidence without ground-truth perception labels, we introduce an intrinsic visual certainty estimation that combines (i) visual grounding measured by KL-divergence under image perturbations and (ii) internal certainty measured by token entropy. We further propose token-level advantage reweighting to focus optimization on tokens based on visual certainty, suppressing ungrounded hallucinations while preserving valid perception. Experiments on thirteen benchmarks show that VL-Calibration effectively improves calibration while boosting visual reasoning accuracy, and it generalizes to out-of-distribution benchmarks across model scales and architectures.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Visual Reasoning and Calibration on Aggregate of 12 evaluation benchmarks Average](<https://www.sota2.com/research/sota/visual-reasoning-and-calibration-on-aggregate-of-12-evaluation-benchmarks-average.md>): Visual Reasoning and Calibration; Aggregate of 12 evaluation benchmarks Average; Accuracy: 80.3.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.09529>)

[PDF](<https://arxiv.org/pdf/2604.09529.pdf>)
