Multimodal Question Answering on MM-Lifelong @day (test)
99.2AccuracyHuman
Evaluation results
| Method | Links | ||
|---|---|---|---|
| HumanFrames=Full2026.03 | 99.2 | 49.8 | |
| ReMAFrames=Full2026.03 | 16.75 | 11.51 | |
| GPT-5Frames=502026.03 | 15.25 | 0.53 | |
| Qwen3-VL-235B-A22BFrames=15362026.03 | 12.44 | 0.79 | |
| Qwen3-VL-30B-A3BFrames=15362026.03 | 11.48 | 0.42 | |
| DeepVideoDiscoveryFrames=Full2026.03 | 10.25 | 3.04 | |
| Video-XL-2-8BFrames=10242026.03 | 9 | 0.72 | |
| Video-XL-2-8BFrames=20482026.03 | 8.75 | 1.37 | |
| Eagle-2.5-8BFrames=322026.03 | 8.25 | 0.39 | |
| VideoMind-7BFrames=Full2026.03 | 7.5 | 1.12 | |
| Eagle-2.5-8BFrames=5122026.03 | 7.25 | 1.01 | |
| Nemotron-v2-12BFrames=5122026.03 | 7.25 | 0.04 | |
| Nemotron-v2-12BFrames=1282026.03 | 7 | 0.03 | |
| LongVT-7BFrames=Full2026.03 | 7 | 0.73 |