Step 3.7 Flash step-3.7-flash

stepfun-ai · stepfun-ai

unranked — no quality signals yet

params
201.4B
arch
moe
context
256k
license
apache-2.0 open weights
released
May 2026
downloads/30d
12.0k

# architecture

attn shared MHA ×64 router top-8 of 288 expert ×288 expert out ×45 layers ctx 262,144 201.4B pool
layers
45
d_model
4096
heads
64q / 64kv
head dim
128
experts
top-8 of 288
vocab
128896
family
step3p5

signal path of one layer, generated from the registry's structured fields — dimension lines quote the model's real numbers; an MoE trace forks at the router, a dense trace runs straight through. Geometry fields come from the repo's config.json.

# trend

OGM score · last 47 days
OGM score over 47 days
overall rank (up is better)
Overall rank over 47 days

# where to run

providerquantctx$/M in$/M out$/M cacheprice srctpsuptime✓
Featherless — ——— via hfrouter — —
DeepInfra 256k $0.20$1.15— via hfrouter 26 —
Novitafp8 256k $0.20$1.15— novita — 100.0% ✓
StepFunfp8 250k $0.20$1.15$0.04 via openrouter — 100.0%
Kyma 250k $0.22$1.28— via litellm — —
Aihubmix 250k $0.22$1.32— via litellm — —

sorted by blended price ((3·input + output) / 4 per 1M) · ✓ = the provider's own catalog confirms the offer · "via …" prices are what the aggregator routing the offer charges, not the provider's own list price

source aliases
openrouter
stepfun/step-3.7-flash