Model comparison

DreamOmni2 vs FLUX.1-dev

At Q4_K_M, DreamOmni2 is the smaller download — 4,683,071,936 bytes against 6,931,817,760.

From the file· summed bytes, KV per layer

Side by side

DreamOmni2FLUX.1-dev
Parameters7.6B11.9B
Architectureqwen2vlflux
Layers
Native context
Mixture of expertsnono
Quantizations published1237
Smallest quantization2.81 GiB3.75 GiB
Q4_K_M4.36 GiB6.46 GiB
Licenceapache-2.0other

KV cache by context

the term that decides long-context viability
ContextDreamOmni2FLUX.1-devRatio
4,096
8,192
16,384
32,768
65,536
131,072