# VLM Judges Can Rank but Cannot Score: Task-Dependent Uncertainty in Multimodal Evaluation

arXiv: https://arxiv.org/abs/2604.25235 · HTML: https://mlsanity.com/paper/2604.25235/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 35 of its 40 references that are in the map.
- First posted on arXiv 28 Apr 2026.

## Placed in

- Multimodal LLMs / vision-language models / [MLLM training and visual instruction tuning](https://mlsanity.com/mllm/mllm-training-and-visual-instruction-tuning/)

## Builds on (most cited first, 30 of 35)

- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment](https://mlsanity.com/paper/2303.16634/) (2023)
- [Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMs](https://mlsanity.com/paper/2306.13063/) (2023)
- [Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedback](https://mlsanity.com/paper/2305.14975/) (2023)
- [Conformalized Quantile Regression](https://mlsanity.com/paper/1905.03222/) (2019)
- [MLLM-as-a-Judge: Assessing Multimodal LLM-as-a-Judge with Vision-Language Benchmark](https://mlsanity.com/paper/2402.04788/) (2024)
- [Learn then Test: Calibrating Predictive Algorithms to Achieve Risk Control](https://mlsanity.com/paper/2110.01052/) (2021)
- [Conformal Language Modeling](https://mlsanity.com/paper/2306.10193/) (2023)
- [Phi-4-reasoning Technical Report](https://mlsanity.com/paper/2504.21318/) (2025)
- [Conformal Prediction with Large Language Models for Multi-Choice Question Answering](https://mlsanity.com/paper/2305.18404/) (2023)
- [LLaVA-Critic: Learning to Evaluate Multimodal Models](https://mlsanity.com/paper/2410.02712/) (2024)
- [Language Models with Conformal Factuality Guarantees](https://mlsanity.com/paper/2402.10978/) (2024)
- [VL-RewardBench: A Challenging Benchmark for Vision-Language Generative Reward Models](https://mlsanity.com/paper/2411.17451/) (2024)
- [Conformal Prediction using Conditional Histograms](https://mlsanity.com/paper/2105.08747/) (2021)
- [A comparison of some conformal quantile regression methods](https://mlsanity.com/paper/1909.05433/) (2019)
- [Fair Conformal Predictors for Applications in Medical Imaging](https://mlsanity.com/paper/2109.04392/) (2021)
- [API Is Enough: Conformal Prediction for Large Language Models Without Logit-Access](https://mlsanity.com/paper/2403.01216/) (2024)
- [Polos: Multimodal Metric Learning from Human Feedback for Image Captioning](https://mlsanity.com/paper/2402.18091/) (2024)
- [SocREval: Large Language Models with the Socratic Method for Reference-Free Reasoning Evaluation](https://mlsanity.com/paper/2310.00074/) (2023)
- [Boosted Conformal Prediction Intervals](https://mlsanity.com/paper/2406.07449/) (2024)
- [Analyzing Uncertainty of LLM-as-a-Judge: Interval Evaluations with Conformal Prediction](https://mlsanity.com/paper/2509.18658/) (2025)
- [Locally Valid and Discriminative Prediction Intervals for Deep Learning Models](https://mlsanity.com/paper/2106.00225/) (2021)
- [Learning Conformal Abstention Policies for Adaptive Risk Management in Large Language and Vision-Language Models](https://mlsanity.com/paper/2502.06884/) (2025)
- [Conformal Prediction via Regression-as-Classification](https://mlsanity.com/paper/2404.08168/) (2024)
- [Black-box Uncertainty Quantification Method for LLM-as-a-Judge](https://mlsanity.com/paper/2410.11594/) (2024)
- [TRACER: Trajectory Risk Aggregation for Critical Episodes in Agentic Reasoning](https://mlsanity.com/paper/2602.11409/) (2026)
- [Learnable Conformal Prediction with Context-Aware Nonconformity Functions for Robotic Planning and Perception](https://mlsanity.com/paper/2509.21955/) (2025)
- [Navigating the Unknown: Uncertainty-Aware Compute-in-Memory Autonomy of Edge Robotics](https://mlsanity.com/paper/2401.17481/) (2024)
- [Beyond Confidence: Adaptive Abstention in Dual-Threshold Conformal Prediction for Autonomous System Perception](https://mlsanity.com/paper/2502.07255/) (2025)
- [Calibrated Decomposition of Aleatoric and Epistemic Uncertainty in Deep Features for Inference-Time Adaptation](https://mlsanity.com/paper/2511.12389/) (2025)

## Built on by (most cited first, 8 of 8)

- [PaintBench: Deterministic Evaluation of Precise Visual Editing](https://mlsanity.com/paper/2606.00188/) (2026)
- [Uncertainty Quantification for Computer-Use Agents: A Benchmark across Vision-Language Models and GUI Grounding Datasets](https://mlsanity.com/paper/2606.25760/) (2026)
- [MIBE: Multi-subject Interaction Benchmark and Evaluator for Personalized Image Generation](https://mlsanity.com/paper/2607.01383/) (2026)
- [CURA: Certified Runtime Alarms for Computer-Use Agents](https://mlsanity.com/paper/2608.27808/) (2026)
- [QVal: Cheaply Evaluating Dense Supervision Signals for Long-Horizon LLM Agents](https://mlsanity.com/paper/2606.32034/) (2026)
- [REMEDY: How Far Is Video Generation from Medical Education World Models?](https://mlsanity.com/paper/2609.32460/) (2026)
- [A Low Grounding Score Is Not an Ungrounded Judge: Identifying the Perceptibility Confound in Multimodal Oversight](https://mlsanity.com/paper/2610.00111/) (2026)
- [Beyond Plausibility: Verifiable Fine-Grained Image Editing on Structured Assets](https://mlsanity.com/paper/2610.04381/) (2026)

## Abstract (arXiv)

Vision-language models (VLMs) are increasingly used as automated judges for multimodal systems, yet their scores provide no indication of reliability. We study this problem through conformal prediction, a distribution-free framework that converts a judge's point score into a calibrated prediction interval using only score-token log-probabilities, with no retraining. We present the first systematic analysis of conformal prediction for VLM-as-a-Judge across 3 judges and 14 visual task categories. Our results show that evaluation uncertainty is strongly task-dependent: intervals cover ~40% of the score range for aesthetics and natural images but expand to ~70% for chart and mathematical reasoning, yielding a quantitative reliability map for multimodal evaluation. We further identify a failure mode not captured by standard evaluation metrics, ranking-scoring decoupling, where judges achieve high ranking correlation while producing wide, uninformative intervals, correctly ordering responses but failing to assign reliable absolute scores. Finally, we show that interval width is driven primarily by task difficulty and annotation quality, i.e., the same judge and method yield 4.5x narrower intervals on a clean, multi-annotator captioning benchmark. Code: https://github.com/divake/VLM-Judge-Uncertainty