AesEval-Bench - Region Selection: leaderboard
Metric: Accuracy (%) of choosing, among four candidates (the flawed element's box, two other element boxes and None), the region with the aesthetic issue for the focal indicator, averaged over AesEval-Bench's 12 indicators; input is the task question with the indicator's explanation, the design image and its layout, font and color metadata as JSON; higher is better. Source: arxiv.org. Saturation forecast: Around May 2028. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 69.89 | #91 |
| 2 | GPT-4o | 67.45 | #333 |
| 3 | Qwen 2.5 VL 72B Instruct | 66.26 | #364 |
| 4 | O3 | 65.81 | #121 |
| 5 | InternVL3-14B | 63.78 | #494 |
| 6 | O1 | 63.47 | #177 |
| 7 | Qwen 2.5 VL 32B Instruct | 63.11 | #443 |
| 8 | Gemini 2.5 Pro | 61 | #145 |
| 9 | InternVL3-8B | 57.99 | #606 |
| 10 | Qwen 2.5 VL 7B Instruct | 57.95 | #643 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=aeseval-bench-region-selection · How It Works · Data refreshed daily, snapshot 2026-10-11.