MUSE - Cultural Identification: leaderboard
Identify the cultural context of an artistic image, MUSE's culture-understanding task; the image set deliberately centres Singaporean and Southeast Asian traditions alongside Western art. Accuracy, higher is better. Source: arxiv.org/abs/2609.19088.
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Around June 2027. 30 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol | 76.5 | #22 |
| 2 | Gemma 3 12B (IT) | 55 | #672 |
| 3 | Qwen 2.5 VL 7B | 54.5 | #641 |
| 4 | InternVL3-38B | 51 | #413 |
| 5 | InternVL3-8B | 49 | #624 |
| 6 | Gemma 3 27B (IT) | 48 | #528 |
| 7 | InternVL3-14B | 46.5 | #506 |
| 8 | Gemma 3 4B (IT) | 43.5 | #997 |
| 9 | MiniCPM-V-2.6 | 43 | #827 |
| 10 | GPT-4o | 39.5 | #357 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=muse-cultural-identification · How It Works · Data refreshed daily, snapshot 2026-11-05.