M3R-Bench - Metaphor Detection: leaderboard
M3R-Bench (Jiang et al., 2026) contains 1,000 human-annotated image-text pairs from social media and advertisements for multimodal metaphor understanding, where each model jointly predicts metaphor presence, target and source concepts and sentiment. This board scores metaphor detection by macro-F1 in percent; higher is better.
Metric: Macro-F1 (%; whether an image-text pair contains a multimodal metaphor, 1,000 instances, 510 metaphorical, predicted jointly with target, source and sentiment). Source: arxiv.org. Saturation forecast: Around 2030. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 32B Instruct | 61.82 | #582 |
| 2 | Kimi K2.5 | 55.12 | #252 |
| 3 | GPT-5.5 | 54.56 | #48 |
| 4 | Qwen 3 VL 8B Instruct | 52.26 | #982 |
| 5 | GPT-4o | 50.19 | #741 |
| 6 | Claude Sonnet 4.6 | 39.03 | #158 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=m3r-bench-metaphor-detection · How It Works · Data refreshed daily, snapshot 2026-11-12.