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

Qwen3.5-4B

by Qwen

4.7B params · image-text-to-text · 669 likes · 9.6M downloads

Qwen3.5-4B is a 4.7B parameter model. At Q4 quantization it requires 2GB of VRAM. It runs comfortably on GeForce RTX 4090 (281 tok/s), GeForce RTX 5090 (422 tok/s), M4 Max 128GB (103 tok/s).

Inference providers

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

GPU compatibility

GPUVRAMQ4 DecodeVerdict
GeForce RTX 409024GB281 tok/scomfortable
GeForce RTX 509032GB422 tok/scomfortable
M4 Max 128GB128GB103 tok/scomfortable
M4 Pro 48GB48GB51 tok/scomfortable
M4 Pro 24GB24GB51 tok/scomfortable
A100 PCIe 80 GB80GB516 tok/scomfortable
H100 SXM5 80 GB80GB1038 tok/scomfortable
GeForce RTX 309024GB249 tok/scomfortable
Radeon RX 7900 XTX24GB206 tok/scomfortable
GeForce RTX 408016GB199 tok/scomfortable
Install CLI [email protected] Raw data · MIT · API data: live · HW/Cloud data: curated 2026-02-23 · v0.6.0