MUSE - Emotion Cause Inference: leaderboard
Infer why an artistic image evokes the emotion it does, one of MUSE's three affective-interpretation tasks. Score, higher is better. Source: arxiv.org/abs/2609.19088.
Metric: Score (%). Source: arxiv.org. Saturation forecast: Not forecast. 30 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol | 49.18 | #22 |
| 2 | Gemma 3 27B (IT) | 35.74 | #528 |
| 3 | Gemma 3 12B (IT) | 33.66 | #672 |
| 4 | InternVL3-38B | 30.95 | #413 |
| 5 | Gemma 3 4B (IT) | 29.36 | #997 |
| 6 | MiniCPM-V-2.6 | 28.79 | #827 |
| 7 | Qwen 2.5 VL 7B | 28.77 | #641 |
| 8 | InternVL3-8B | 27.98 | #624 |
| 9 | GPT-4o | 27.97 | #357 |
| 10 | InternVL3-14B | 27.59 | #506 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=muse-emotion-cause-inference · How It Works · Data refreshed daily, snapshot 2026-11-05.