MoHallBench - Scene-Action Hallucination (Multiple Choice): leaderboard
Metric: Multiple-choice accuracy (%; four-option multiple choice (ground truth, hard hallucination distractor, random distractor, None of these); 32 uniformly sampled frames, temperature 0; scene-action co-occurrence hallucination). Source: arxiv.org. Saturation forecast: Around December 2026. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Molmo2-8B | 78.2 |
| 2 | Qwen 3 VL 8B Instruct | 76.1 |
| 3 | InternVL3-8B | 74.5 |
| 4 | Qwen 2.5 VL 7B Instruct | 72.8 |
| 5 | Qwen 2 VL 7B Instruct | 71 |
Interactive version: theaggregate.ai/benchmark?slug=mohallbench-scene-action-hallucination-multiple-choice · How It Works · Data refreshed daily, snapshot 2026-09-29.