Multimodal Hallucination - AVH Visual RefCheck Resistance: leaderboard
The "AVH Visual RefCheck Resistance" subtask of Multimodal Hallucination. A task measuring hallucination in multimodal models — how often a vision-language model asserts objects, attributes or facts not supported by the image.
Metric: Resistance (100 - RefCheck). Source: huggingface.co. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini-v1.5-pro (audio-only & audio-visual) | 80.66 | |
| 2 | Gemini-v1.5-pro (visual-only) | 75.81 | |
| 3 | Chat-Univi | 74.29 | |
| 4 | Video-LLaVA | 68.29 | |
| 5 | Llama-VID | 64.44 | |
| 6 | FAVOR (visual-only) | 54.7 | |
| 7 | Video-Llama (visual-only) | 53.66 | |
| 8 | Video-Llama (audio-only & audio-visual) | 49.59 | |
| 9 | FAVOR (audio-only & audio-visual) | 48.58 | |
| 10 | Valley | 48.22 |
Interactive version: theaggregate.ai/benchmark?slug=multimodal-hallucination-avh-visual-refcheck-resistance · How It Works · Data refreshed daily, snapshot 2026-10-20.