Model comparison

Anima vs Krea-2-Raw

At Q4_K_M, Anima is the smaller download — 1,384,407,168 bytes against 7,486,289,184.

From the file· summed bytes, KV per layer

Side by side

AnimaKrea-2-Raw
Parameters2.1B12.8B
Architecturecosmoskrea2
Layers
Native context
Mixture of expertsnono
Quantizations published2329
Smallest quantization0.95 GiB2.34 GiB
Q4_K_M1.29 GiB6.97 GiB
Licenceotherother

KV cache by context

the term that decides long-context viability
ContextAnimaKrea-2-RawRatio
4,096
8,192
16,384
32,768
65,536
131,072