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

Step-3.5-Flash

by stepfun-ai

199B params · text-generation · 701 likes · 132.7k downloads

Step-3.5-Flash is a 199B parameter model. At Q4 quantization it requires 100GB of VRAM. It runs comfortably on H200 SXM 141 GB (35 tok/s).

Inference providers

Provider$/1M in$/1M outThroughput
Featherless

GPU compatibility

GPUVRAMQ4 DecodeVerdict
M4 Max 128GB128GB2 tok/stight
H200 SXM 141 GB141GB35 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