MuseBench - Static Visual Arts (Multi-Select): leaderboard
Metric: Static-visual-arts exact-match accuracy (%) on multi-select questions; higher is better. Source: arxiv.org. Saturation forecast: Around 2038. 28 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 | 30.38 |
| 2 | InternVL3-8B | 27.69 |
| 3 | GPT-5.4 | 27.15 |
| 4 | Qwen 3.5 397B A17B | 23.12 |
| 5 | Qwen 3.5 Plus | 18.82 |
| 6 | InternVL3-78B | 18.01 |
| 7 | Qwen2.5-Omni-7B | 17.47 |
| 8 | Gemini 3.1 Pro (Preview) | 16.4 |
| 9 | Gemma 4 E4B | 13.17 |
| 10 | Grok 4.1 | 10.48 |
| 11 | GLM-4.5V | 9.14 |
| 12 | Kimi K2.5 | 2.69 |
Interactive version: theaggregate.ai/benchmark?slug=musebench-static-visual-arts-multi-select · How It Works · Data refreshed daily, snapshot 2026-09-29.