Text-to-Video generation on Wan2.1 14B CFG = 5.0, 720 × 1280p, frames = 80 (test) benchmark leaderboard · SOTA2 Research

Text-to-Video generation on Wan2.1 14B CFG = 5.0, 720 × 1280p, frames = 80 (test)

0.182CLIPSIMFull Prec.

Evaluation Results

Method	Links
Full Prec.2025.09		0.182	90.79	0	—	—	—
QuantSparse2025.09		0.182	89.96	0.002	18.67	0.622	0.24
QuantSparse2025.09		0.181	92.87	0.06	18.67	0.616	0.277
QuaRot2025.09		0.18	66.56	0.313	13.59	0.409	0.566
ViDiT-Q2025.09		0.18	71.26	0.251	15.3	0.513	0.376
Q-VDiT2025.09		0.18	89.1	0.082	18.13	0.61	0.264
QuaRot+SVG2025.09		0.18	60.14	0.396	13.55	0.399	0.567
QuaRot+SVG2025.09		0.179	67.64	0.336	13.6	0.407	0.555
Q-VDiT+SVG2025.09		0.179	88.29	0.091	16.69	0.563	0.323
Q-VDiT+SVG2025.09		0.179	85.26	0.182	15.94	0.532	0.367
SmoothQuant2025.09		0.178	62.25	0.363	13.06	0.404	0.656