Visual Aesthetic Benchmark: Can Frontier Models Judge Beauty?

TL;DR

This paper introduces the Visual Aesthetic Benchmark (VAB) to evaluate whether multimodal large language models can accurately judge visual beauty through comparative selection, revealing a significant gap compared to expert human judgment.

Contribution

The paper presents VAB, a new benchmark for aesthetic evaluation based on comparative selection, and evaluates frontier models, highlighting their limitations and the benefits of fine-tuning.

Findings

Current models identify best/worst images correctly only 26.5% of the time, below human performance of 68.9%.

Fine-tuning improves model accuracy, approaching human-level performance.

VAB exposes the gap between AI models and expert aesthetic judgment, providing a new evaluation standard.

Abstract

Multimodal large language models (MLLMs) are now routinely deployed for visual understanding, generation, and curation. A substantial fraction of these applications require an explicit aesthetic judgment. Most existing solutions reduce this judgment to predicting a scalar score for a single image. We first ask whether such scores faithfully capture comparative preference: in a controlled study with eight expert annotators, score-derived rankings align poorly with the same annotators' direct comparisons, while direct ranking yields substantially higher inter-annotator agreement on best- and worst-image labels. Motivated by this finding, we introduce the Visual Aesthetic Benchmark (VAB), which casts aesthetic evaluation as comparative selection over candidate sets with matched subject matter. VAB contains 400 tasks and 1,195 images across fine art, photography, and illustration, with labels derived from the consensus of 10 independent expert judges per task. Evaluating 20 frontier MLLMs and six dedicated visual-quality reward models, we find that the strongest system identifies both the best and the worst image correctly across three random permutations of the candidate order in only 26.5% of tasks, far below the 68.9% achieved by human experts. Fine-tuning a 35B-parameter model on 2,000 expert examples brings its accuracy close to that of a 397B-parameter open-weight model, suggesting that the comparative signal in VAB is transferable. Together, these results expose a clear and measurable gap between current multimodal models and expert aesthetic judgment, and VAB provides the first set-based, expert-grounded testbed on which that gap can be tracked and closed.