Visual Aesthetic Benchmark (VAB) — leaderboard
Metric: Overall Top-1 ap@1 (self-reported). Source: benchmarklist.com. 22 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Human Expert | 77.7 |
| 2 | Claude Sonnet 4.6 | 54.8 |
| 3 | Gemini 3.1 Pro (Preview) | 54.8 |
| 4 | Claude Opus 4.6 | 53 |
| 5 | Gemini 3 Flash (Preview) | 52.8 |
| 6 | GPT-5.1 | 52.3 |
| 7 | Seed 2.0 Pro | 51.3 |
| 8 | Claude Opus 4.5 | 51.2 |
| 9 | GPT-5 | 51 |
| 10 | GPT-5.1 Codex | 50.2 |
| 11 | Qwen 3.5 397B A17B | 50.1 |
| 12 | GPT-4.1 | 49.3 |
| 13 | O4 Mini | 48.2 |
| 14 | Claude Sonnet 4.5 | 47.2 |
| 15 | GPT-5.2 | 46.3 |
Interactive version: theaggregate.ai/benchmark?slug=visual-aesthetic-benchmark-vab · How the rankings work · Data refreshed daily, snapshot 2026-07-22.