List your company

Video-LLM Performance Calibration on Eval-OOD (EgoSchema & LongVideoBench)

5.8ECEVideoABC
4.6412.4720.328.13May 30, 2026

Evaluation results

MethodLinks
VideoABCTarget Video-LLM=LLaVA-OV (7B), Model Size Category=Large size2026.055.8—
VideoABCModel Size=–, Target Video-LLM=Qwen-3.5-VL (9B)2026.058.7226
VideoABCTarget Video-LLM=Qwen-3.5-VL (9B), Model Size Category=Large size2026.058.7—
VideoABCTarget Video-LLM=Video-LLaVA 7B, Model Size Category=Large size2026.059.7—
VideoABCTarget Video-LLM=LLaVA-Next 7B, Model Size Category=Large size2026.0510.2—
JudgeTarget Video-LLM=LLaVA-Next 7B, Model Size Category=Large size, Judge Model=LLaVA-OV (7B)2026.0512.1—
MLP-ATarget Video-LLM=LLaVA-Next 7B, Model Size Category=Large size2026.0514.6—
JudgeTarget Video-LLM=LLaVA-OV (7B), Model Size Category=Large size, Judge Model=LLaVA-OV (7B)2026.0514.8—
MLP-ATarget Video-LLM=LLaVA-OV (7B), Model Size Category=Large size2026.0515.6—
VideoABCTarget Video-LLM=Qwen-3.5-VL 4B, Model Size Category=Small size2026.0515.6—
JudgeModel Size=72B, Target Video-LLM=Qwen-3.5-VL (9B)2026.0516.41,802
JudgeModel Size=7B, Target Video-LLM=Qwen-3.5-VL (9B)2026.0517.1656
JudgeTarget Video-LLM=Qwen-3.5-VL (9B), Model Size Category=Large size, Judge Model=LLaVA-OV (7B)2026.0517.1—
VideoABCTarget Video-LLM=LLaVA-OV 0.5B, Model Size Category=Small size2026.0517.2—
JudgeTarget Video-LLM=Video-LLaVA 7B, Model Size Category=Large size, Judge Model=LLaVA-OV (7B)2026.0517.3—
MLP-AModel Size=–, Target Video-LLM=Qwen-3.5-VL (9B)2026.0518.2119
JudgeModel Size=0.5B, Target Video-LLM=Qwen-3.5-VL (9B)2026.0518.7167
JudgeModel Size=3B, Target Video-LLM=Qwen-3.5-VL (9B)2026.0518.7427
DirectTarget Video-LLM=Video-LLaVA 7B, Model Size Category=Large size2026.0518.9—
MLP-ATarget Video-LLM=Video-LLaVA 7B, Model Size Category=Large size2026.0519.2—
DirectModel Size=–, Target Video-LLM=Qwen-3.5-VL (9B)2026.0519.4201
DirectTarget Video-LLM=Qwen-3.5-VL (9B), Model Size Category=Large size2026.0519.4—
MLP-ATarget Video-LLM=Qwen-3.5-VL (9B), Model Size Category=Large size2026.0519.7—
MLP-ATarget Video-LLM=Qwen-3.5-VL 4B, Model Size Category=Small size2026.0519.8—
JudgeTarget Video-LLM=LLaVA-OV 0.5B, Model Size Category=Small size, Judge Model=LLaVA-OV (7B)2026.0519.8—
DirectTarget Video-LLM=LLaVA-OV (7B), Model Size Category=Large size2026.0520—
DirectTarget Video-LLM=LLaVA-Next 7B, Model Size Category=Large size2026.0520.3—
JudgeTarget Video-LLM=Qwen-3.5-VL 4B, Model Size Category=Small size, Judge Model=LLaVA-OV (7B)2026.0520.8—
MLP-2LModel Size=9B, Target Video-LLM=Qwen-3.5-VL (9B)2026.0521.7678
MLP-2LTarget Video-LLM=Qwen-3.5-VL (9B), Model Size Category=Large size2026.0521.7—
MLP-ATarget Video-LLM=LLaVA-OV 0.5B, Model Size Category=Small size2026.0523.4—
DirectTarget Video-LLM=Qwen-3.5-VL 4B, Model Size Category=Small size2026.0524.6—
DirectTarget Video-LLM=LLaVA-OV 0.5B, Model Size Category=Small size2026.0525.6—
MLP-2LTarget Video-LLM=LLaVA-OV (7B), Model Size Category=Large size2026.0528.5—
MLP-2LTarget Video-LLM=Video-LLaVA 7B, Model Size Category=Large size2026.0529.2—
RandomModel Size=-, Target Video-LLM=Qwen-3.5-VL (9B)2026.0529.310
RandomTarget Video-LLM=Qwen-3.5-VL (9B), Model Size Category=Large size2026.0529.3—
MLP-2LTarget Video-LLM=LLaVA-Next 7B, Model Size Category=Large size2026.0530.8—
MLP-2LTarget Video-LLM=Qwen-3.5-VL 4B, Model Size Category=Small size2026.0532.2—
RandomTarget Video-LLM=LLaVA-OV (7B), Model Size Category=Large size2026.0532.8—
RandomTarget Video-LLM=LLaVA-Next 7B, Model Size Category=Large size2026.0532.9—
RandomTarget Video-LLM=Video-LLaVA 7B, Model Size Category=Large size2026.0534.6—
MLP-2LTarget Video-LLM=LLaVA-OV 0.5B, Model Size Category=Small size2026.0534.7—
RandomTarget Video-LLM=Qwen-3.5-VL 4B, Model Size Category=Small size2026.0534.8—
RandomTarget Video-LLM=LLaVA-OV 0.5B, Model Size Category=Small size2026.0534.8—