List your company

Video Question Answering on OVO-Bench Real-Time Visual Perception

91.95OCRQwen2.5-VL-7B + SAVEMem
4.69427.3475072.653May 8, 2026

Evaluation results

MethodLinks
Qwen2.5-VL-7B + SAVEMemSize=7B, Frames=1 fps, Mode=Online, Training Strategy=Training-Free2026.0591.9568.8181.0365.7370.371.7474.93
Qwen2.5-VL-3B + SAVEMemSize=3B, Frames=1 fps, Mode=Online, Training Strategy=Training-Free2026.0589.2668.8173.2855.6264.3669.5770.15
Gemini 1.5 ProSize=-, Frames=1 fps, Mode=Proprietary2026.0585.9166.9779.3158.4363.3761.9669.32
ThinkStreamSize=3B, Frames=1 fps, Mode=Online, Training Strategy=Training-Based2026.0585.2364.2269.8349.4469.3164.1367.03
Qwen2.5-VL-3BSize=3B, Frames=1 fps, Mode=Online, Training Strategy=Training-Free2026.0577.1852.2968.9741.0167.3360.8761.27
ViSpeakSize=7B, Frames=1 fps, Mode=Online, Training Strategy=Training-Based2026.0575.258.7271.5551.1274.2666.8566.3
TimeChat-OnlineSize=7B, Frames=1 fps, Mode=Online, Training Strategy=Training-Based2026.0575.246.870.747.869.361.461.9
GPT-4oSize=-, Frames=64, Mode=Proprietary2026.0569.864.2271.5551.1270.359.7864.46
LLaVA-VideoSize=7B, Frames=64, Mode=Offline2026.0569.859.6366.3850.5672.2861.4163.34
Qwen2-VLSize=7B, Frames=64, Mode=Offline2026.0569.1353.2163.7950.5666.3460.8760.65
InternVL2Size=8B, Frames=64, Mode=Offline2026.0568.4658.7268.9744.9467.3355.9860.73
StreamForestSize=7B, Frames=1 fps, Mode=Online, Training Strategy=Training-Based2026.0568.4653.2171.5547.7565.3560.8761.2
Qwen2.5-VL-7BSize=7B, Frames=1 fps, Mode=Online, Training Strategy=Training-Free2026.0567.7955.0567.2442.1366.3460.8759.9
DispiderSize=7B, Frames=1 fps, Mode=Online, Training Strategy=Training-Based2026.0557.7249.5462.0744.9461.3951.6354.55
LongVUSize=7B, Frames=1 fps, Mode=Offline2026.0555.749.5459.4848.3168.3263.0457.4
Flash-VStreamSize=7B, Frames=1 fps, Mode=Online, Training Strategy=Training-Based2026.0525.532.1129.3133.7129.728.829.86
VideoLLM-OnlineSize=8B, Frames=2 fps, Mode=Online, Training Strategy=Training-Based2026.058.0523.8512.0714.0445.5421.220.79