Model comparison

HunyuanVideo-I2V vs Wan2.2-Animate-14B

At Q4_K_M, HunyuanVideo-I2V is the smaller download — 7,883,680,512 bytes against 11,496,331,072.

From the file· summed bytes, KV per layer

Side by side

HunyuanVideo-I2VWan2.2-Animate-14B
Parameters12.8B17.3B
Architecturehyvidwan
Layers
Native context
Mixture of expertsnono
Quantizations published2417
Smallest quantization2.96 GiB6.01 GiB
Q4_K_M7.34 GiB10.71 GiB
Licenceotherapache-2.0

KV cache by context

the term that decides long-context viability
ContextHunyuanVideo-I2VWan2.2-Animate-14BRatio
4,096
8,192
16,384
32,768
65,536
131,072