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

dolphin-2.5-mixtral-8x7b

by dphn

46.7B params · text-generation · 1.2k likes · 1.6k downloads

dolphin-2.5-mixtral-8x7b is a 46.7B parameter model. At Q4 quantization it requires 23GB of VRAM. It runs comfortably on GeForce RTX 5090 (42 tok/s), A100 PCIe 80 GB (51 tok/s), H100 SXM5 80 GB (103 tok/s).

GPU compatibility

GPUVRAMQ4 DecodeVerdict
GeForce RTX 509032GB42 tok/scomfortable
M4 Max 128GB128GB10 tok/stight
M4 Pro 48GB48GB5 tok/stight
A100 PCIe 80 GB80GB51 tok/scomfortable
H100 SXM5 80 GB80GB103 tok/scomfortable
M4 Max 64GB64GB10 tok/stight
H200 SXM 141 GB141GB150 tok/scomfortable
M2 Ultra 192GB192GB14 tok/stight
Install CLI [email protected] Raw data · MIT · API data: live · HW/Cloud data: curated 2026-02-23 · v0.6.0