Model comparison

Qwen-Image vs FLUX.1-dev

At Q4_K_M, FLUX.1-dev is the smaller download — 6,931,817,760 bytes against 13,065,746,976.

From the file· summed bytes, KV per layer

Side by side

Qwen-ImageFLUX.1-dev
Parameters20.4B11.9B
Architectureqwen_imageflux
Layers
Native context
Mixture of expertsnono
Quantizations published4337
Smallest quantization6.58 GiB3.75 GiB
Q4_K_M12.17 GiB6.46 GiB
Licenceapache-2.0other

KV cache by context

the term that decides long-context viability
ContextQwen-ImageFLUX.1-devRatio
4,096
8,192
16,384
32,768
65,536
131,072