OVO-Bench: How Far is Your Video-LLMs from Real-World Online Video Understanding?

Benchmark Model Rank Results
video-question-answering-on-ovo-benchDispider-7B–Real-Time Visual Perception: 54.55Backward Tracing: 36.06…
video-question-answering-on-ovo-benchFlash-VStream-7B–Real-Time Visual Perception: 28.37Backward Tracing: 27.38…
video-question-answering-on-ovo-benchGPT-4o–Real-Time Visual Perception: 64.46Backward Tracing: 60.75…
video-question-answering-on-ovo-benchGemini 1.5 Pro–Real-Time Visual Perception: 69.32Backward Tracing: 62.54…
video-question-answering-on-ovo-benchInternVL2-8B–Real-Time Visual Perception: 60.39Backward Tracing: 43.44…
video-question-answering-on-ovo-benchLLaVA-OneVision-7B–Real-Time Visual Perception: 64.02Backward Tracing: 43.71…
video-question-answering-on-ovo-benchLLaVA-Video-7B–Real-Time Visual Perception: 63.52Backward Tracing: 40.4…
video-question-answering-on-ovo-benchLongVU-7B–Real-Time Visual Perception: 57.61Backward Tracing: 35.01…
video-question-answering-on-ovo-benchQwen2-VL-72B–Real-Time Visual Perception: 61.92Backward Tracing: 56.95…
video-question-answering-on-ovo-benchQwen2-VL-7B–Real-Time Visual Perception: 55.98Backward Tracing: 46.46…
video-question-answering-on-ovo-benchVideoLLM-online-8B–Real-Time Visual Perception: 20.79Backward Tracing: 17.73