Video Understanding on VideoMME Long
59ScoreDIG
Evaluation Results
| Method | Links | |
|---|---|---|
| DIG#Frames=768, Backbone=Qwen3-VL-8B2025.12 | 59 | |
| AKS#Frames=768, Backbone=Qwen3-VL-8B2025.12 | 57.3 | |
| UNI#Frames=768, Backbone=Qwen3-VL-8B2025.12 | 56.6 | |
| DIG#Frames=512, Backbone=Qwen3-VL-8B2025.12 | 56.4 | |
| AKS#Frames=512, Backbone=Qwen3-VL-8B2025.12 | 56 | |
| UNI#Frames=512, Backbone=Qwen3-VL-8B2025.12 | 55 | |
| AKS#Frames=192, Backbone=Qwen3-VL-8B2025.12 | 53.8 | |
| DIG#Frames=256, Backbone=Qwen3-VL-8B2025.12 | 53.8 | |
| UNI#Frames=256, Backbone=Qwen3-VL-8B2025.12 | 53.4 | |
| AKS#Frames=256, Backbone=Qwen3-VL-8B2025.12 | 53.2 | |
| AKS#Frames=64, Backbone=Qwen3-VL-8B2025.12 | 51.6 | |
| UNI#Frames=192, Backbone=Qwen3-VL-8B2025.12 | 51.6 | |
| UNI#Frames=128, Backbone=Qwen3-VL-8B2025.12 | 51.3 | |
| AKS#Frames=128, Backbone=Qwen3-VL-8B2025.12 | 51.1 | |
| DIG#Frames=128, Backbone=Qwen3-VL-8B2025.12 | 51.1 | |
| DIG#Frames=192, Backbone=Qwen3-VL-8B2025.12 | 50.7 | |
| AKS#Frames=32, Backbone=Qwen3-VL-8B2025.12 | 50.1 | |
| DIG#Frames=16, Backbone=Qwen3-VL-8B2025.12 | 49.9 | |
| DIG#Frames=8, Backbone=Qwen3-VL-8B2025.12 | 49.8 | |
| DIG#Frames=32, Backbone=Qwen3-VL-8B2025.12 | 49.4 | |
| UNI#Frames=8, Backbone=Qwen3-VL-8B2025.12 | 49.2 | |
| DIG#Frames=64, Backbone=Qwen3-VL-8B2025.12 | 49 | |
| UNI#Frames=64, Backbone=Qwen3-VL-8B2025.12 | 48.9 | |
| UNI#Frames=32, Backbone=Qwen3-VL-8B2025.12 | 48.6 | |
| UNI#Frames=16, Backbone=Qwen3-VL-8B2025.12 | 48 |