Multimodal Understanding on Opencompass Image Benchmark (val)
84MMBench AccuracySAIL-VL1.6-8B
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| SAIL-VL1.6-8B2026.04 | 84 | 69.5 | 55.4 | 74.2 | 90.5 | 87.5 | 54.4 | 73.3 | 73.6 | |
| Ovis2-8B2026.04 | 83.6 | 64.6 | 57.4 | 71.8 | 89.1 | 86.6 | 56.3 | 65.1 | 71.8 | |
| InternVL2.5-8B2026.04 | 82.5 | 63.2 | 56.2 | 64.5 | 82.1 | 84.6 | 49 | 62.8 | 68.1 | |
| Qwen2.5-VL-7B2026.04 | 82.2 | 64.1 | 58 | 68.1 | 88.8 | 84.3 | 51.9 | 69.7 | 70.9 | |
| InternVL3-8B2026.04 | 82.1 | 68.7 | 62.2 | 70.5 | 88.4 | 85.1 | 49 | 82.8 | 73.6 | |
| POINTS-LongEvaluation Mode=focus2026.04 | 82.1 | 66.1 | 53.7 | 70.6 | 85.5 | 84.2 | 48.3 | 66.7 | 69.7 | |
| POINTS-LongEvaluation Mode=focus, Token Pruning/Compression Method=Attn-prune, Compression Ratio=50%2026.04 | 82 | 64.5 | 52.4 | 69 | 83.5 | 84.2 | 47.2 | 66.7 | 68.7 | |
| POINTS1.5-8BEvaluation Mode=baseline2026.04 | 81.9 | 65.7 | 53.2 | 70.9 | 85.8 | 83.9 | 50.1 | 64.7 | 69.5 | |
| POINTS-LongEvaluation Mode=focus, Token Pruning/Compression Method=Folder, Compression Ratio=50%2026.04 | 81.5 | 64 | 52.7 | 64 | 75.5 | 83.5 | 49.1 | 64.9 | 66.9 | |
| Qwen2-VL-7B2026.04 | 81 | 60.7 | 53.7 | 61.6 | 84.3 | 83 | 50.4 | 61.8 | 67.1 | |
| POINTS-LongEvaluation Mode=focus, Token Pruning/Compression Method=Avg-pooling, Compression Ratio=50%2026.04 | 80.7 | 62.2 | 53 | 64.6 | 75.1 | 83.8 | 48.5 | 65.5 | 66.7 | |
| MiniCPM-o-2.62026.04 | 80.6 | 63.3 | 50.9 | 73.3 | 88.9 | 86.1 | 51.1 | 67.2 | 70.2 |