Head to head

RTX 3090 vs Mac Studio M4 Max 128GB

The Mac Studio M4 Max 128GB holds more — 128 GB against 24 GB — which decides what you can load at all. The RTX 3090 has more bandwidth at 936 GB/s, which decides how fast tokens come out once a model fits.

Model (Q4_K_M, 8K)RTX 3090tok/sMac Studio M4 Max 128GBtok/s
Llama 3.1 8B Runs comfortably 112 Runs comfortably 67.8
Llama 3.1 70B Won't fit 1.67 Runs comfortably 8.48
Llama 3.2 3B Runs comfortably 246 Runs comfortably 150
Llama 3.2 1B Runs comfortably 650 Runs comfortably 400
Llama 4 Scout 109B-A17B Won't fit 5.39 Runs comfortably 28.7
Qwen3 8B Runs comfortably 108 Runs comfortably 65.6
Qwen3 14B Runs comfortably 63.5 Runs comfortably 38.3
Qwen3 32B Fits, but tight 29.5 Runs comfortably 17.8
Qwen3 4B Runs comfortably 196 Runs comfortably 120
Qwen3 30B-A3B Runs comfortably 117 Runs comfortably 86.4
Qwen3 235B-A22B Won't fit 3.36 Won't fit 7.42
Qwen2.5-Coder 32B Fits, but tight 29.5 Runs comfortably 17.8
Qwen2.5 7B Runs comfortably 125 Runs comfortably 75.3
Qwen2.5 72B Won't fit 1.59 Runs comfortably 8.24
Gemma 3 4B Runs comfortably 209 Runs comfortably 128
Gemma 3 12B Runs comfortably 76.1 Runs comfortably 46.1

Specifications

RTX 3090Mac Studio M4 Max 128GB
Memory24 GB128 GB
Bandwidth936 GB/s546 GB/s
FP16 compute71 TFLOPS34 TFLOPS
Launch price$1,499$3,499
ArchitectureAmpereM4