Model comparison

Wan2.1_T2V_14B_FusionX_VACE vs Wan2.2-T2V-A14B

At Q4_K_M, Wan2.2-T2V-A14B is the smaller download — 9,650,090,496 bytes against 11,629,623,072.

From the file· summed bytes, KV per layer

Side by side

Wan2.1_T2V_14B_FusionX_VACEWan2.2-T2V-A14B
Parameters17.3B14.3B
Architecturewanwan
Layers
Native context
Mixture of expertsnono
Quantizations published1528
Smallest quantization5.92 GiB4.94 GiB
Q4_K_M10.83 GiB8.99 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextWan2.1_T2V_14B_FusionX_VACEWan2.2-T2V-A14BRatio
4,096
8,192
16,384
32,768
65,536
131,072