List your company

Long Video Understanding on VideoMME (test)

64.2AccuracyNVILA-8B
52.13655.26858.461.532May 21, 2026

Evaluation results

MethodLinks
NVILA-8BBackbone=8B2026.0564.2
LLaVA-Video-7B + ST-GridPoolInput frames=64, Pooling Strategy=ST-GridPool2026.0564.2
LLaVA-Video-7BInput frames=642026.0563.3
Apollo-7BBackbone=7B2026.0561.3
LLaVA-OneVision-7B + ST-GridPoolInput frames=32, Pooling Strategy=ST-GridPool2026.0559
Oryx-1.5-7BBackbone=1.5B2026.0558.8
LLaVA-OneVision-7BInput frames=322026.0558.2
IXC-2.5-7BBackbone=7B2026.0555.8
VideoLLaMA2.1-7BBackbone=7B2026.0554.9
mPLUG-Owl3-8BBackbone=8B2026.0553.5
LongVA-7BBackbone=7B2026.0552.6