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

Qwen3.5-27B

by Qwen

27.8B params · image-text-to-text · 638 likes · 835.0k downloads

Qwen3.5-27B is a 27.8B parameter model. At Q4 quantization it requires 14GB of VRAM. It runs comfortably on GeForce RTX 4090 (47 tok/s), GeForce RTX 5090 (70 tok/s), A100 PCIe 80 GB (86 tok/s).

Inference providers

Provider$/1M in$/1M outThroughput
Novita58 tok/s

GPU compatibility

GPUVRAMQ4 DecodeVerdict
GeForce RTX 409024GB47 tok/scomfortable
GeForce RTX 509032GB70 tok/scomfortable
M4 Max 128GB128GB17 tok/stight
M4 Pro 48GB48GB8 tok/stight
M4 Pro 24GB24GB8 tok/stight
A100 PCIe 80 GB80GB86 tok/scomfortable
H100 SXM5 80 GB80GB174 tok/scomfortable
GeForce RTX 309024GB41 tok/scomfortable
Radeon RX 7900 XTX24GB34 tok/scomfortable
M4 Max 64GB64GB17 tok/stight
Install CLI [email protected] Raw data · MIT · API data: live · HW/Cloud data: curated 2026-02-23 · v0.6.0