List your company

Video Summarization on Video-MME Long Videos

44.6Clip CoherenceGemini 2.5 Pro
34.40837.05439.742.346Jan 8, 2026

Evaluation results

MethodLinks
Gemini 2.5 Protau=0.6, # Segments number of clips=9.21 ± 10.40, Clips meaningful text length=325.6, Speed (Generation time per clips)=120s avg / video, Cost in Dollar=1.25 input / 10 output per M Token2026.0144.667.441.5
Gemini 2.5 Flashtau=0.6, # Segments number of clips=18.32 ± 31.03, Clips meaningful text length=203.5, Speed (Generation time per clips)=80–85s avg / video, Cost in Dollar=0.3 input / 2.5 output per M Token2026.014170.137.9
Infinite Video-to-Video Clips Generationtau=0.6, # Segments number of clips=6.46 ± 1.67, Clips meaningful text length=258.3, Speed (Generation time per clips)=30–55s / video, Cost in Dollar=0.3 input / 2.5 output per M Token2026.0134.872.133.9