Video Understanding on VideoMME (Accuracy)
71.9AccuracyGPT-4o
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-4o2026.05 | 71.9 | |
| Qwen3-VLArchitecture Type=Modular, Parameter Scale=8B, Training Stage=Instruct2026.05 | 71.4 | |
| NEO-ovArchitecture Type=Native, Parameter Scale=8B, Training Stage=Instruct2026.05 | 67.4 | |
| VideoLLaMA3Architecture Type=Modular, Parameter Scale=8B, Training Stage=Instruct2026.05 | 66.2 | |
| InternVL3.5Architecture Type=Modular, Parameter Scale=8B, Training Stage=Instruct2026.05 | 66 | |
| Qwen2.5-VL-7B + SAGESAGE=true2026.05 | 65.1 | |
| LLaVA-VideoArchitecture Type=Modular, Parameter Scale=8B, Training Stage=Instruct2026.05 | 63.3 | |
| Video-RTS-7B2026.05 | 63 | |
| Qwen3-VLArchitecture Type=Modular, Parameter Scale=2B, Training Stage=Instruct2026.05 | 61.9 | |
| ViSS-R1-7B2026.05 | 60.5 | |
| NEO-ovArchitecture Type=Native, Parameter Scale=2B, Training Stage=Instruct2026.05 | 60.4 | |
| VILA-1.5-40B2026.05 | 60.1 | |
| VideoRFT-7B2026.05 | 59.8 | |
| VideoLLaMA3Architecture Type=Modular, Parameter Scale=2B, Training Stage=Instruct2026.05 | 59.6 | |
| InternVL3.5Architecture Type=Modular, Parameter Scale=2B, Training Stage=Instruct2026.05 | 58.4 | |
| LLaVA-OneVision-7B2026.05 | 58.2 | |
| Qwen2.5-VL-7B + SFT + SAGESFT=true, SAGE=true2026.05 | 58.2 | |
| Video-R1-7B2026.05 | 57.4 | |
| Kangaroo-8B2026.05 | 56 | |
| VideoLLaMA2 + REGATELLM=Qwen2-7B2025.07 | 54.5 | |
| LLaVA-OneVision + LaViLLM=Qwen2-7B2025.07 | 54 | |
| LongVA-7B2026.05 | 52.6 | |
| Video-UTR-7B2026.05 | 52.6 | |
| Qwen2.5-VL-7B + SFTSFT=true2026.05 | 51.7 | |
| Qwen2.5-VL-7B2026.05 | 50.6 | |
| VideoLLaMA22026.05 | 47.9 | |
| ELVAArchitecture Type=Native, Parameter Scale=8B, Training Stage=Instruct2026.05 | 47.1 | |
| MSA-PTTraining Strategy=from-scratch sparse pretraining2026.06 | 45.48 | |
| ELVAArchitecture Type=Native, Parameter Scale=2B, Training Stage=Instruct2026.05 | 41.8 | |
| FullTraining Strategy=Full-Attention baseline2026.06 | 41.11 | |
| MSA-CPTTraining Strategy=sparse continued pretraining2026.06 | 39.65 | |
| EVEArchitecture Type=Native, Parameter Scale=8B, Training Stage=Instruct2026.05 | 29.3 | |
| FuyuArchitecture Type=Native, Parameter Scale=8B, Training Stage=Instruct2026.05 | 28.7 |