vram.run Models Hardware Providers Cloud State of Inference
API provider data is live · Hardware & cloud pricing curated 2026-02-23

Step-3.7-Flash

by stepfun-ai

201B params · image-text-to-text · 383 likes · 102.0k downloads

Step-3.7-Flash is a 201B parameter model. At Q4 quantization it requires 101GB of VRAM. It runs comfortably on H200 SXM 141 GB (34 tok/s).

Inference providers

Provider$/1M in$/1M outThroughput
Featherless
deepinfra109 tok/s

GPU compatibility

GPUVRAMQ4 DecodeVerdict
M4 Max 128GB128GB2 tok/stight
H200 SXM 141 GB141GB34 tok/scomfortable
M2 Ultra 192GB192GB3 tok/stight
Install CLI [email protected] Raw data · MIT · API data: live · HW/Cloud data: curated 2026-02-23 · v0.6.0