Multimodal Reasoning on InfoVQA
5,511.1Token CountLOOK-M
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LOOK-MRetention level=≤ 5%2026.05 | 5,511.1 | 27.39 | |
| FastVRetention level=≤ 10%2026.05 | 2,963.6 | 20.63 | |
| LOOK-MRetention level=≤ 10%2026.05 | 2,694.1 | 40.94 | |
| VisionzipRetention level=≤ 10%2026.05 | 2,533.3 | 22.66 | |
| FastVRetention level=≤ 5%2026.05 | 2,239.7 | 21.78 | |
| VisionzipRetention level=≤ 5%2026.05 | 1,468.5 | 22.04 | |
| FastV-VPRetention level=≤ 5%2026.05 | 665 | 81.9 | |
| BaselineRetention level=100%2026.05 | 623.1 | 84.37 | |
| FastV-VPRetention level=≤ 10%2026.05 | 530.7 | 83.62 |