Pest Classification on QFSD (test)
85.44Beetle AccuracyGPT-5
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| GPT-5Model Type=Proprietary2026.05 | 85.44 | 92.59 | 89.32 | 96.67 | 62.5 | 54.55 | 87.29 | 64.71 | 87.5 | 83.28 | |
| Gemini-2.5-FlashModel Type=Proprietary2026.05 | 83.95 | 84.21 | 87.82 | 96.15 | 78.08 | 70 | 89.16 | 75 | 40 | 84.94 | |
| Pest-ThinkerParameter Count=7B, Model Type=Our Model2026.05 | 81.26 | 92.59 | 86.12 | 96.67 | 87.75 | 90.91 | 84.37 | 84.12 | 87.5 | 84.32 | |
| GPT-5-miniModel Type=Proprietary2026.05 | 78.64 | 74.07 | 73.79 | 90 | 51.04 | 36.36 | 83.05 | 47.06 | 50 | 72.56 | |
| Pest-ThinkerParameter Count=3B, Model Type=Our Model2026.05 | 76.55 | 82.59 | 88.83 | 80 | 73.33 | 81.82 | 78.14 | 52.94 | 62.5 | 75.96 | |
| Pest-Thinker-SFTParameter Count=7B, Model Type=Our Model2026.05 | 72.52 | 74.07 | 77.86 | 83.33 | 78.54 | 74.62 | 77.29 | 41.18 | 75 | 75.33 | |
| Qwen2.5VL-InstructParameter Count=72B, Model Type=Open-source Vanilla2026.05 | 71.84 | 51.85 | 60.68 | 73.33 | 51.04 | 54.55 | 66.95 | 52.94 | 62.5 | 62.46 | |
| Pest-Thinker-SFTParameter Count=3B, Model Type=Our Model2026.05 | 71.84 | 62.96 | 65.73 | 80 | 65.62 | 72.73 | 66.27 | 47.06 | 62.5 | 63.71 | |
| Qwen3VL-ThinkingParameter Count=8B, Model Type=Open-source Reasoning2026.05 | 61.62 | 37.5 | 51.87 | 5 | 41.94 | 63.64 | 63.21 | 46.15 | 42.86 | 54.91 | |
| Qwen3VL-ThinkingParameter Count=30B, Model Type=Open-source Reasoning2026.05 | 61.29 | 56 | 63.21 | 77.78 | 57.14 | 22.22 | 73.45 | 71.43 | 25 | 63.39 | |
| VisionThink-EfficientParameter Count=7B, Model Type=Open-source Reasoning2026.05 | 57.28 | 37.04 | 57.28 | 76.67 | 46.88 | 54.55 | 65.25 | 41.18 | 12.5 | 56.31 | |
| VisionThink-GeneralParameter Count=7B, Model Type=Open-source Reasoning2026.05 | 55.34 | 51.85 | 55.83 | 83.33 | 44.79 | 54.55 | 66.95 | 47.06 | 0 | 56.15 | |
| Vision-G1Parameter Count=7B, Model Type=Open-source Reasoning2026.05 | 55.34 | 48.15 | 54.85 | 80 | 47.92 | 45.45 | 59.32 | 52.94 | 0 | 54.42 | |
| Qwen2.5VL-InstructParameter Count=32B, Model Type=Open-source Vanilla2026.05 | 54.37 | 48.15 | 43.69 | 86.67 | 50 | 54.55 | 62.71 | 70.59 | 25 | 52.68 | |
| Qwen3VL-InstructParameter Count=30B, Model Type=Open-source Vanilla2026.05 | 54 | 65.38 | 51.96 | 80 | 45.74 | 18.18 | 70.34 | 76.47 | 75 | 57.03 | |
| Qwen2.5VL-InstructParameter Count=7B, Model Type=Open-source Vanilla2026.05 | 48.54 | 25.93 | 48.54 | 76.67 | 44.79 | 27.27 | 59.32 | 29.41 | 0 | 49.21 | |
| Vision-SR1Parameter Count=7B, Model Type=Open-source Reasoning2026.05 | 47.57 | 29.63 | 47.57 | 66.67 | 44.79 | 27.27 | 42.37 | 47.06 | 12.5 | 46.06 | |
| Qwen3VL-InstructParameter Count=8B, Model Type=Open-source Vanilla2026.05 | 47.52 | 43.48 | 34.95 | 70.37 | 46.81 | 27.27 | 55.26 | 40 | 62.5 | 44.89 | |
| Qwen2.5VL-InstructParameter Count=3B, Model Type=Open-source Vanilla2026.05 | 45.63 | 25.93 | 35.44 | 76.67 | 45.83 | 9.09 | 56.78 | 41.18 | 25 | 43.85 | |
| InternVL3Parameter Count=8B, Model Type=Open-source Vanilla2026.05 | 42.72 | 48.15 | 40.78 | 53.33 | 46.88 | 0 | 49.15 | 29.41 | 0 | 42.74 | |
| InternVL3.5Parameter Count=8B, Model Type=Open-source Vanilla2026.05 | 36.89 | 25.93 | 39.32 | 40 | 36.36 | 54.55 | 39.83 | 52.94 | 25 | 38.96 |