MoHallBench - Visual Similarity Hallucination: leaderboard
Metric: Q-PairAcc (%; share of adversarial pairs in which the model answers both the positive and the negative (reversed) yes/no query correctly; 32 uniformly sampled frames, temperature 0; visually near-identical negative clips without the action). Source: arxiv.org. Saturation forecast: Around December 2026. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 VL 8B Instruct | 36.2 |
| 2 | InternVL3-8B | 27.9 |
| 3 | Molmo2-8B | 24.4 |
| 4 | Qwen 2.5 VL 7B Instruct | 22.3 |
| 5 | Qwen 2 VL 7B Instruct | 17.4 |
Interactive version: theaggregate.ai/benchmark?slug=mohallbench-visual-similarity-hallucination · How It Works · Data refreshed daily, snapshot 2026-09-29.