Anomaly Discrimination on Anomaly Discrimination Tasks
95.24AccuracyHuman (expert)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Human (expert)Scale=-2026.02 | 95.24 | 94.25 | 98.89 | 96.43 | |
| Human (ordinary)Scale=-2026.02 | 86.9 | 87.07 | 94.35 | 89.3 | |
| Reason-IAD (Qwen3-VL)Scale=8B2026.02 | 79.42 | 73.77 | 83.37 | 73.08 | |
| Reason-IAD (Qwen3-VL)Scale=4B2026.02 | 79.2 | 68.21 | 86.4 | 75.49 | |
| Qwen3-VLScale=8B2026.02 | 74.51 | 75.58 | 79.76 | 77.43 | |
| Qwen3-VLScale=4B2026.02 | 74.48 | 57.95 | 89.41 | 67.01 | |
| Reason-IAD (Qwen2.5-VL)Scale=7B2026.02 | 73.91 | 55.5 | 75.95 | 62.15 | |
| Reason-IAD (Qwen3-VL)Scale=2B2026.02 | 73.48 | 92.23 | 69.19 | 78.75 | |
| Qwen3-VLScale=2B2026.02 | 69.24 | 66.83 | 79.78 | 71.64 | |
| Qwen2.5-VLScale=7B2026.02 | 68.87 | 53.74 | 75.54 | 61.48 | |
| Gemini-1.5-proScale=-2026.02 | 68.63 | 45.47 | 86.84 | 57.6 | |
| GPT-4oScale=-2026.02 | 68.63 | 67.37 | 75.68 | 71.04 | |
| InternVL2Scale=76B2026.02 | 68.25 | 55.81 | 83.52 | 64.4 | |
| AnomalyGPTScale=7B2026.02 | 65.57 | 82.11 | 74.45 | 76.68 | |
| GPT-4o-miniScale=-2026.02 | 64.33 | 65.47 | 73.04 | 68.67 | |
| claude-3.5-sonnetScale=-2026.02 | 60.14 | 30.87 | 76.75 | 41.92 | |
| InternVL2Scale=8B2026.02 | 59.97 | 30.25 | 79.22 | 41.23 | |
| Gemini-1.5-flashScale=-2026.02 | 58.58 | 78.63 | 67.41 | 72.4 | |
| LLaVA-NeXTScale=34B2026.02 | 57.92 | 46.27 | 69.98 | 54.44 | |
| LLaVA-NeXT-InterleaveScale=7B2026.02 | 57.64 | 16.58 | 90.83 | 25.64 | |
| MiniCPM-V2.6Scale=8B2026.02 | 57.31 | 34.38 | 70.98 | 45.31 | |
| InternLM-XComposer2-VLScale=7B2026.02 | 55.85 | 17.94 | 75.87 | 27.16 | |
| Cambrian-1*Scale=8B2026.02 | 55.6 | 22.28 | 74.1 | 31.85 | |
| Qwen-VL-ChatScale=7B2026.02 | 53.65 | 43.95 | 65.39 | 47.28 | |
| SPHINX*Scale=7B2026.02 | 53.13 | 6.42 | 99.74 | 10.61 | |
| LLaVA-OnVisionScale=7B2026.02 | 51.77 | 4.9 | 78.19 | 9.1 | |
| LLaVA-1.5Scale=7B2026.02 | 51.33 | 94.79 | 62.72 | 75.32 | |
| LLaVA-1.5Scale=13B2026.02 | 49.96 | 99.79 | 62 | 76.28 |