howtospark
models / devstral-small-2
params
24.0B
sparsity
Dense
attention
Full attention
architecture
mistral3
license
apache-2.0
downloads
384.8k
likes
628
hf updated
Jun 30, 2026

Visualize

total size

each column is a layer — every parameter fires on every token

bits / param
16-bit
footprint
44.7 GiB
sparks to fit
block ≈
188M params
active paramsghost = precision traded awayblock fill ∝ bits kept · Devstral Small 224.01136184B params

Eval scores (compare all)

Reported

68SWE-bench Verified