ArtECulture: leaderboard
Metric: Accuracy (%; mean of the English, Chinese and Arabic culture accuracies; zero-shot prediction of the majority emotion (nine categories) that the target culture perceives in each of the 679 test artworks; original models without the authors retrieval knowledge or fine-tuning). Source: arxiv.org. Saturation forecast: Around 2033. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.5 Flash | 49.88 |
| 2 | Gemini 3 Flash | 49.68 |
| 3 | Qwen 3.5 27B | 48.4 |
| 4 | Gemma 4 31B (IT) | 48.01 |
| 5 | Claude Opus 4.8 | 47.86 |
| 6 | Claude Sonnet 4.6 | 46.59 |
| 7 | Qwen 3.6 35B A3B | 46.05 |
| 8 | Qwen 3.6 27B | 44.77 |
| 9 | GPT-5.5 | 44.13 |
| 10 | Qwen 3.5 9B | 43.45 |
| 11 | Qwen 3 VL 8B | 41.58 |
| 12 | Gemma 3 12B (IT) | 39.37 |
| 13 | aya-vision-8B | 39.13 |
| 14 | Pixtral-12B | 32.5 |
Interactive version: theaggregate.ai/benchmark?slug=arteculture · How It Works · Data refreshed daily, snapshot 2026-09-26.