List your company

Real-Time Visual Perception on OVO-Bench

94OCRHuman Agents
4.61227.818551.02574.2315Mar 13, 2026Mar 22, 2026Mar 31, 2026Apr 10, 2026Apr 19, 2026Apr 28, 2026May 8, 2026

Evaluation results

MethodLinks
Human Agents2026.039492.694.892.791.19493.2—
Human2026.059492.694.892.791.19493.292.8
Response-G1Params=8B2026.0590.674.375.959.669.371.773.661.3
ThinkStreamModel Category=Open-source Online Models, Model Scale=7B2026.0387.9259.637553.9370.3—69.1264.9
Gemini 1.5 Pro# Frames=1fps2026.0387.36780.254.568.367.470.8—
Gemini 1.5 Pro2026.0587.36780.254.568.367.470.865.3
ThinkStreamModel Category=Open-source Online Models, Model Scale=3B2026.0385.2364.2269.8249.4369.3164.1367.0359.66
StreamoModel Category=Open-source Online Models, Model Scale=3B2026.0378.5252.2967.2444.3855.4571.261.5151.64
Qwen2.5-VLModel Category=Open-source Offline Models, Model Scale=32B2026.0377.1858.7268.150.5674.2657.6164.457.37
Qwen2.5-VLModel Category=Open-source Offline Models, Model Scale=3B2026.0376.5144.0367.2442.1368.3161.9660.0351
Em-Garde-7B# Frames=1fps2026.0376.561.565.547.270.357.163—
StreamAgent-7B# Frames=1fps2026.0371.253.263.653.967.358.761.3—
StreamAgentParams=7B2026.0571.253.263.653.967.358.761.349.4
LLaVA-Video-7B# Frames=642026.0369.859.666.450.672.361.463.3—
LLaVA-NeXT-VideoParams=7B2026.0569.859.666.450.672.361.463.353.1
TimeChat-OnlineParams=7B2026.0569.848.664.744.968.355.458.645.6
LLaVA-VideoModel Category=Open-source Offline Models, Model Scale=7B2026.0369.1358.7268.8349.4474.2659.7863.5251.96
GPT-4o# Frames=642026.0369.165.165.55068.363.763.6—
Qwen2-VL-7B# Frames=642026.0369.153.263.850.666.360.960.7—
GPT-4o2026.0569.165.165.55068.363.763.658.6
Qwen2-VLParams=7B2026.0569.153.263.850.666.360.960.752.7
InternVL-V2-8B# Frames=642026.0368.558.76944.967.35660.7—
StreamForest-7B# Frames=1fps2026.0368.553.271.647.865.460.961.2—
InternVL-V2Params=8B2026.0568.558.76944.967.35660.750.1
StreamForestModel Category=Open-source Online Models, Model Scale=7B2026.0368.4653.2171.5547.7565.3560.8761.256.61
Qwen2.5-VLModel Category=Open-source Offline Models, Model Scale=7B2026.0367.7955.0567.2442.1366.3460.8759.952.28
LLaVA-OneVision-7B# Frames=642026.0367.158.769.849.471.360.362.8—
LLaVA-OneVisionParams=7B2026.0567.158.769.849.471.360.362.852.9
Qwen2-VLModel Category=Open-source Offline Models, Model Scale=7B2026.0360.450.4656.0347.1966.3455.4355.9851.22
DispiderModel Category=Open-source Online Models, Model Scale=7B2026.0357.7249.5462.0744.9461.3951.6354.5545.31
Dispider# Frames=1fps2026.0357.749.562.144.961.451.654.5—
DispiderParams=7B2026.0557.749.562.144.961.451.654.541.8
LongVU-7B# Frames=1fps2026.0355.749.559.548.368.36357.4—
LongVUParams=7B2026.0555.749.559.548.368.36357.448.5
LongVUModel Category=Open-source Offline Models, Model Scale=7B2026.0353.6953.2162.9347.7568.3259.7857.6146.31
Flash-VStream-7B# Frames=1fps2026.0325.532.129.333.729.728.829.9—
Flash-VstreamParams=7B2026.0525.532.129.333.729.728.829.933.2
Flash-VStreamModel Category=Open-source Online Models, Model Scale=7B2026.0324.1629.3628.4533.7125.7428.828.3727.88
VideoLLM-online-8B# Frames=2fps2026.038.123.912.11445.521.220.8—
VideoLLM-onlineParams=8B2026.058.123.912.11445.521.220.8—
VideoLLM-onlineModel Category=Open-source Online Models, Model Scale=8B2026.038.0523.8512.0714.0445.5421.220.7919.26