Sens-VisualNews: leaderboard
Sens-VisualNews (2026) tests whether multimodal models can tell if a news image is sensational, meaning built to provoke shock, fear, anger or disgust, over 9,576 VisualNews images balanced between the two classes. Models answer yes or no zero-shot, and the score is top-1 accuracy in percent on the test split (chance 50); higher is better.
Metric: Top-1 accuracy (%) on the test split of the full dataset (majority-vote labels of three annotators), zero-shot yes or no answer to whether the news image is sensational, greedy decoding, each model with the prompt and prompt format that scored best on the development split; balanced classes, chance 50; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 12 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | LLaVA-OneVision-1.5-8B | 87.6 | #1864 |
| 2 | Qwen 3 VL 8B | 87 | #917 |
| 3 | LLaVA-OneVision-1.5-4B | 86.7 | #1855 |
| 4 | Qwen 3 VL 2B | 86.3 | #2134 |
| 5 | Qwen 3 VL 4B | 85 | #1151 |
| 6 | InternVL3.5-8B | 83.8 | #1604 |
| 7 | InternVL3.5-4B | 83.4 | #1930 |
| 8 | LLaVA-OneVision-7B | 81.5 | #2981 |
| 9 | LLaVA-OneVision-0.5B | 81.3 | #7956 |
| 10 | InternVL3.5-2B | 77.9 | #3284 |
| 11 | InternVL3.5-1B | 76.1 | #6112 |
| 12 | SmolVLM2 2.2B | 62.2 |
Interactive version: theaggregate.ai/benchmark?slug=sens-visualnews · How It Works · Data refreshed daily, snapshot 2026-11-21.