AI-generated image detection (test)
0.751Macro F1 ScoreGPT-5.4
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| GPT-5.4Parameters=≈1.5T2026.06 | 0.751 | — | — | — | — | — | — | |
| Yuvion VL-32BParameters=32B2026.06 | 0.741 | — | — | — | — | — | — | |
| Yuvion VL-8BParameters=8B2026.06 | 0.736 | — | — | — | — | — | — | |
| Qwen3.5-PlusParameters=397B2026.06 | 0.696 | — | — | — | — | — | — | |
| K2.5Parameters=≈1T2026.06 | 0.658 | — | — | — | — | — | — | |
| Qwen3-VL-32BParameters=32B2026.06 | 0.604 | — | — | — | — | — | — | |
| Qwen3-VL-8BParameters=8B2026.06 | 0.558 | — | — | — | — | — | — | |
| GLASSBackbone Model=ViT-Base/162026.01 | — | 97.533 | 0.997 | 0.975 | 0.975 | 0.975 | 0.058 | |
| GLASSBackbone Model=ConvNeXt-Tiny2026.01 | — | 98.611 | 0.998 | 0.986 | 0.986 | 0.986 | 0.048 | |
| GLASSBackbone Model=ResNet-502026.01 | — | 97.656 | 0.997 | 0.977 | 0.977 | 0.977 | 0.052 | |
| StandardBackbone Model=ViT-Base/162026.01 | — | 93.244 | 0.972 | 0.932 | 0.933 | 0.932 | 0.033 | |
| StandardBackbone Model=ConvNeXt-Tiny2026.01 | — | 96.133 | 0.99 | 0.961 | 0.961 | 0.961 | 0.041 | |
| StandardBackbone Model=ResNet-502026.01 | — | 93.656 | 0.981 | 0.937 | 0.937 | 0.937 | 0.04 |