List your company

Long-video Question Answering on Long VideoBench (val)

73.2AccuracyGOPAgen
37.73646.94356.1565.357Jan 14, 2025Apr 8, 2025Jul 1, 2025Sep 23, 2025Dec 16, 2025Mar 10, 2026Jun 3, 2026

Evaluation results

MethodLinks
GOPAgenType=Agentic video understanding framework2026.0673.2
GPT-5LLM=N/A, # Frames=N/A2025.1272.6
Deep Video DiscoveryType=Agentic video understanding framework2026.0671.6
VideoLucyType=Agentic video understanding framework2026.0670.2
OpenAI o3Access=Closed-Source2026.0667.5
AdaReTaKeAccess=Open-Source2026.0667
GPT-4o2025.0166.7
GPT-4oFrames=2562025.0266.7
GPT-4oLLM=N/A, # Frames=N/A2025.1266.7
GPT-4oAccess=Closed-Source2026.0666.7
Gemini-1.5-Pro2025.0164
Gemini-1.5-ProFrames=2562025.0264
Gemini-1.5-ProLLM=N/A, # Frames=N/A2025.1264
Gemini-1.5-ProAccess=Closed-Source2026.0664
InternVL2.5-78B2025.0163.6
InternVL2.5-78BAccess=Open-Source2026.0663.6
LLaVA-Video + OneClip-RAGLLM=7B, # Frames=642025.1262.9
LLaVA-Video w/ AKSFrames=64, LLM=7B, sampling=AKS2025.0262.7
AKSLLM=7B, # Frames=642025.1262.7
InternVL3.5LLM=8B, # Frames=642025.1262.1
LLaVA-Video-72B2025.0161.9
Gemini-1.5-FlashFrames=2562025.0261.6
GPT-4VFrames=2562025.0261.3
Qwen2.5-VL-72BAccess=Open-Source2026.0660.7
Qwen2-VL w/ AKSFrames=32, LLM=7B, sampling=AKS2025.0260.5
InternVL2.5-8B2025.0160
VideoNSALLM=7B, # Frames=20482025.1260
mPLUG-Owl3Access=Open-Source2026.0659.8
LLaVA-OV w/ AKSFrames=32, LLM=7B, sampling=AKS2025.0259.3
LLaVA-VideoLLM=7B, # Frames=642025.1259.3
LLaVA-VideoFrames=64, LLM=7B2025.0258.9
Tarsier2-7BNumber of frames=128f2025.0158.6
LLaVA-Video-7B2025.0158.2
NVILALLM=8B, # Frames=2562025.1257.7
Qwen2.5-VLLLM=7B, # Frames=322025.1256
Qwen2-VL-7B2025.0155.6
Qwen2-VLFrames=32, LLM=7B2025.0255.5
MiniCPM-V 2.6Frames=64, LLM=8B2025.0254.9
LLaVA-OVFrames=32, LLM=7B2025.0254.8
Tarsier-34B2025.0154.2
PLLAVAFrames=32, LLM=34B2025.0253.2
mPLUG-Owl3LLM=7B, # Frames=162025.1252.1
Video-XLLLM=7B, # Frames=1282025.1250.7
LongVALLM=7B, # Frames=2562025.1247.8
Tarsier-7B2025.0139.8
VideoLLaVAFrames=8, LLM=7B2025.0239.1