Multimodal Deception Evaluation on MM-DeceptionBench primary 1.0
92.41PrecisionGPT-4o
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| GPT-4oClass=Deceptive2025.11 | 92.41 | 50.12 | 64.99 | — | — | — | — | |
| Gemini-2.5-ProClass=Deceptive2025.11 | 85.23 | 85.02 | 85.13 | — | — | — | — | |
| Claude-Sonnet-4Class=Deceptive2025.11 | 83.17 | 83.37 | 83.27 | — | — | — | — | |
| Gemini-2.5-ProClass=Non-deceptive2025.11 | 62.87 | 63.25 | 63.06 | 35.76 | 14.98 | 78.79 | 0.4819 | |
| Claude-Sonnet-4Class=Non-deceptive2025.11 | 58.18 | 57.83 | 58.01 | 42.17 | 16.63 | 76.08 | 0.4128 | |
| GPT-4oClass=Non-deceptive2025.11 | 41.97 | 89.76 | 57.2 | 10.24 | 49.88 | 61.49 | 0.2975 |