Video Understanding on OVO-Bench (Accuracy)
69.3AccuracyGemini 1.5 Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini 1.5 ProFrames=–2026.06 | 69.3 | |
| Qwen2.5-VL + CausalMemFrames=1fps2026.06 | 67.8 | |
| ViSpeakFrames=1fps2026.06 | 66.3 | |
| LLaVA-OneVision + CausalMemFrames=0.5/0.2fps2026.06 | 65.7 | |
| GPT-4oFrames=–2026.06 | 64.5 | |
| LLaVA-VideoFrames=642026.06 | 63.5 | |
| LLaVA-OneVisionFrames=322026.06 | 62.6 | |
| TimeChat-OnlineFrames=1fps2026.06 | 61.4 | |
| LongVUFrames=1fps2026.06 | 57.4 | |
| DispiderFrames=1fps2026.06 | 54.6 | |
| Flash-VStreamFrames=1fps2026.06 | 28.4 | |
| VideoLLM-OnlineFrames=2fps2026.06 | 20.8 |