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

Qwen3-Next-80B-A3B-Instruct

by Qwen

81.3B params · text-generation · 949 likes · 898.0k downloads

Qwen3-Next-80B-A3B-Instruct is a 81.3B parameter model. At Q4 quantization it requires 41GB of VRAM. It runs comfortably on H100 SXM5 80 GB (59 tok/s), H200 SXM 141 GB (86 tok/s).

Inference providers

Provider$/1M in$/1M outThroughput
Novita101 tok/s
Hyperbolic175 tok/s
Together AI136 tok/s

GPU compatibility

GPUVRAMQ4 DecodeVerdict
M4 Max 128GB128GB5 tok/stight
M4 Pro 48GB48GB2 tok/stight
A100 PCIe 80 GB80GB29 tok/stight
H100 SXM5 80 GB80GB59 tok/scomfortable
M4 Max 64GB64GB5 tok/stight
H200 SXM 141 GB141GB86 tok/scomfortable
M2 Ultra 192GB192GB8 tok/stight
Install CLI [email protected] Raw data · MIT · API data: live · HW/Cloud data: curated 2026-02-23 · v0.6.0