Multimodal Model Evaluation on MME (Score, Accuracy, Relative Performance)
2,369.1MME ScoreBaseline
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| BaselineBackbone=InternVL3-8B, Retained Tokens=All 1280 Tokens, Reduction Percentage=100%2026.06 | 2,369.1 | 74.9 | 100 | |
| TOPSBackbone=InternVL3-8B, Retained Tokens=256 Tokens, Reduction Percentage=↓ 80.0%2026.06 | 2,302 | 72.4 | 96.7 | |
| BaselineBackbone=Qwen2.5-VL-7B-Instruct, Retained Tokens=All 1296 Tokens, Reduction Percentage=100%2026.06 | 2,301.8 | 73.1 | 100 | |
| FastVBackbone=InternVL3-8B, Retained Tokens=256 Tokens, Reduction Percentage=↓ 80.0%2026.06 | 2,289.4 | 71.2 | 95.1 | |
| TOPSBackbone=Qwen2.5-VL-7B-Instruct, Retained Tokens=256 Tokens, Reduction Percentage=↓ 80.2%2026.06 | 2,284.6 | 70.4 | 96.3 | |
| TOPSBackbone=InternVL3-8B, Retained Tokens=128 Tokens, Reduction Percentage=↓ 90.0%2026.06 | 2,252 | 68.8 | 91.9 | |
| CDPrunerBackbone=Qwen2.5-VL-7B-Instruct, Retained Tokens=256 Tokens, Reduction Percentage=↓ 80.2%2026.06 | 2,231.9 | 68.9 | 94.3 | |
| TOPSBackbone=Qwen2.5-VL-7B-Instruct, Retained Tokens=128 Tokens, Reduction Percentage=↓ 90.1%2026.06 | 2,217.3 | 67.9 | 92.9 | |
| DivPruneBackbone=InternVL3-8B, Retained Tokens=256 Tokens, Reduction Percentage=↓ 80.0%2026.06 | 2,178.5 | 69.3 | 92.5 | |
| FastVBackbone=Qwen2.5-VL-7B-Instruct, Retained Tokens=256 Tokens, Reduction Percentage=↓ 80.2%2026.06 | 2,169.3 | 67.8 | 92.7 | |
| DivPruneBackbone=Qwen2.5-VL-7B-Instruct, Retained Tokens=256 Tokens, Reduction Percentage=↓ 80.2%2026.06 | 2,167.3 | 68.8 | 94.1 | |
| VisionZipBackbone=InternVL3-8B, Retained Tokens=256 Tokens, Reduction Percentage=↓ 80.0%2026.06 | 2,148.8 | 67.8 | 90.5 | |
| CDPrunerBackbone=InternVL3-8B, Retained Tokens=256 Tokens, Reduction Percentage=↓ 80.0%2026.06 | 2,130 | 67.4 | 90 | |
| DivPruneBackbone=InternVL3-8B, Retained Tokens=128 Tokens, Reduction Percentage=↓ 90.0%2026.06 | 2,051 | 64.6 | 86.2 | |
| DivPruneBackbone=Qwen2.5-VL-7B-Instruct, Retained Tokens=128 Tokens, Reduction Percentage=↓ 90.1%2026.06 | 2,044.3 | 65.5 | 89.6 | |
| CDPrunerBackbone=Qwen2.5-VL-7B-Instruct, Retained Tokens=128 Tokens, Reduction Percentage=↓ 90.1%2026.06 | 2,033.2 | 65.5 | 89.6 | |
| CDPrunerBackbone=InternVL3-8B, Retained Tokens=128 Tokens, Reduction Percentage=↓ 90.0%2026.06 | 1,952.6 | 62.8 | 83.8 | |
| VisionZipBackbone=InternVL3-8B, Retained Tokens=128 Tokens, Reduction Percentage=↓ 90.0%2026.06 | 1,864.7 | 60.3 | 80.5 | |
| FastVBackbone=InternVL3-8B, Retained Tokens=128 Tokens, Reduction Percentage=↓ 90.0%2026.06 | 1,806.9 | 60.4 | 80.6 | |
| DCLABackbone=Qwen2-VL-2B-Instruct2026.06 | 1,751.4 | — | — | |
| DAMOBackbone=Qwen2-VL-2B-Instruct2026.06 | 1,749.8 | — | — | |
| FADEBackbone=Qwen2-VL-2B-Instruct2026.06 | 1,746.2 | — | — | |
| GreedyBackbone=Qwen2-VL-2B-Instruct2026.06 | 1,745.9 | — | — | |
| VCDBackbone=Qwen2-VL-2B-Instruct2026.06 | 1,743.9 | — | — | |
| PAIBackbone=Qwen2-VL-2B-Instruct2026.06 | 1,735.8 | — | — | |
| FastVBackbone=Qwen2.5-VL-7B-Instruct, Retained Tokens=128 Tokens, Reduction Percentage=↓ 90.1%2026.06 | 1,596.8 | 57 | 78 |