Multi-shot Multi-reference Video Generation Benchmark benchmark leaderboard · SOTA2 Research

Multi-shot Multi-reference Video Generation Benchmark

86.13Avg Score w/o Alignment-FaceSimPoCo

Evaluation Results

Method	Links
PoCo2026.04		86.13	83.46	89.35	92.66	70.12	51.29	98.87	98.46
EchoShot2026.04		83.82	—	87.05	79.81	—	54.34	99.09	98.79
Phantom-14B2026.04		82.32	80.72	86.12	73.24	72.75	55.25	98.74	98.23
Phantom-1.3B2026.04		80	80.07	80.37	71.15	80.37	52.72	98.43	97.35
VACE-14B2026.04		77.46	75.56	69.49	67.3	67.05	52.6	99	98.89
VACE-1.3B2026.04		75.28	71.6	56.02	69.14	53.21	53.89	98.85	98.5