Model comparison

MiniMax-H3 vs Wan2.2-I2V-A14B

At Q4_K_M, Wan2.2-I2V-A14B is the smaller download — 9,651,728,896 bytes against 14,576,977,960.

From the file· summed bytes, KV per layer

Side by side

MiniMax-H3Wan2.2-I2V-A14B
Parameters33.1B14.3B
Architectureqwen3vlwan
Layers
Native context
Mixture of expertsnono
Quantizations published2230
Smallest quantization10.60 GiB4.94 GiB
Q4_K_M13.58 GiB8.99 GiB
Licenceotherapache-2.0

KV cache by context

the term that decides long-context viability
ContextMiniMax-H3Wan2.2-I2V-A14BRatio
4,096
8,192
16,384
32,768
65,536
131,072