Multimodal Question Answering on MM-Lifelong (val@month)
80.4AccuracyHuman
Evaluation results
| Method | Links | ||
|---|---|---|---|
| HumanFrames=Full2026.03 | 80.4 | 33.5 | |
| ReMAFrames=Full2026.03 | 18.62 | 15.46 | |
| GPT-5Frames=502026.03 | 14.87 | 0.44 | |
| Qwen3-VL-235B-A22BFrames=15362026.03 | 14.33 | 0.06 | |
| Qwen3-VL-30B-A3BFrames=15362026.03 | 11.92 | 0.64 | |
| DeepVideoDiscoveryFrames=Full2026.03 | 10.57 | 4.48 | |
| Nemotron-v2-12BFrames=1282026.03 | 10.03 | 0.01 | |
| Nemotron-v2-12BFrames=5122026.03 | 9.63 | 0.02 | |
| Video-XL-2-8BFrames=10242026.03 | 9.07 | 0.75 | |
| Video-XL-2-8BFrames=20482026.03 | 8.91 | 0.4 | |
| VideoMind-7BFrames=Full2026.03 | 8.35 | 0.26 | |
| LongVT-7BFrames=Full2026.03 | 7.54 | 0.11 | |
| Eagle-2.5-8BFrames=322026.03 | 6.1 | 0.01 | |
| Eagle-2.5-8BFrames=5122026.03 | 4.41 | 0.03 |