Multimodal Hallucination - AVH Audio SelfCheckGPT Resistance: leaderboard
The "AVH Audio SelfCheckGPT 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 - SelfCheckGPT). Source: huggingface.co. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | LTU | 78.05 | |
| 2 | Gemini-v1.5-pro (audio-only) | 74.18 | |
| 3 | FAVOR (audio-visual) | 66.75 | |
| 4 | SALMONN | 65.01 | |
| 5 | Qwen-Audio-Chat | 63.43 | #1244 |
| 6 | Gemini-v1.5-pro (audio-visual) | 51.49 | |
| 7 | FAVOR (audio-only) | 50.38 | |
| 8 | Video-Llama (audio-only) | 43.58 | |
| 9 | Video-Llama (audio-visual) | 29.77 |
Interactive version: theaggregate.ai/benchmark?slug=multimodal-hallucination-avh-audio-selfcheckgpt-resistance · How It Works · Data refreshed daily, snapshot 2026-10-20.