Model comparison

ERNIE-Image-Turbo vs Qwen-Image-Edit-2511

At Q4_K_M, ERNIE-Image-Turbo is the smaller download — 5,019,124,416 bytes against 11,657,701,984.

From the file· summed bytes, KV per layer

Side by side

ERNIE-Image-TurboQwen-Image-Edit-2511
Parameters8.0B20.4B
Architecturewanqwen_image
Layers
Native context
Mixture of expertsnono
Quantizations published1936
Smallest quantization2.96 GiB4.13 GiB
Q4_K_M4.67 GiB10.86 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextERNIE-Image-TurboQwen-Image-Edit-2511Ratio
4,096
8,192
16,384
32,768
65,536
131,072