List your company

Long Video Question Answering on Video-MME Long without Subtitles

62.3Overall AccuracyVideo-RAG
44.93249.44153.9558.459Apr 7, 2026

Evaluation results

MethodLinks
Video-RAGBackbone=72B, Native Input=false2026.0462.31.3
VideoStirBackbone=72B, Native Input=true2026.0462.11
TV-RAGBackbone=72B, Native Input=false2026.0461.80.5
LLaVA-VideoBackbone=72B, Native Input=true2026.0461.5—
Video-RAGBackbone=25B, Native Input=false2026.0459.61.4
VideoStirBackbone=25B, Native Input=true2026.0459.41
TV-RAGBackbone=25B, Native Input=false2026.04590.3
AriaBackbone=25B, Native Input=true2026.0458.8—
VideoStirBackbone=8B, Native Input=true2026.04511.8
TV-RAGBackbone=8B, Native Input=false2026.0450.91.6
Video-RAGBackbone=8B, Native Input=false2026.0450.61
mPLUG-Owl3Backbone=8B, Native Input=true2026.0450.1—
VideoStirBackbone=26B, Native Input=true2026.0446.92.9
Video-RAGBackbone=26B, Native Input=false2026.0446.72.4
TV-RAGBackbone=26B, Native Input=false2026.0446.31.5
InternVL-1.5Backbone=26B, Native Input=true2026.0445.6—