Model comparison

SPARK.Chroma_preview vs FLUX.1-dev

These two publish different quantization sets; the table below has the exact sizes.

From the file· summed bytes, KV per layer

Side by side

SPARK.Chroma_previewFLUX.1-dev
Parameters8.9B11.9B
Architectureflux
Layers
Native context
Mixture of expertsnono
Quantizations published537
Smallest quantization4.71 GiB3.75 GiB
Q4_K_M6.46 GiB
Licenceapache-2.0other

KV cache by context

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