Head to head

RTX 3090 vs RTX 4070 Ti Super

The RTX 3090 holds more — 24 GB against 16 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/sRTX 4070 Ti Supertok/s
Llama 3.1 8B Runs comfortably 112 Runs comfortably 77.6
Llama 3.1 70B Won't fit 1.67 Won't fit 1.22
Llama 3.2 3B Runs comfortably 246 Runs comfortably 172
Llama 3.2 1B Runs comfortably 650 Runs comfortably 456
Llama 4 Scout 109B-A17B Won't fit 5.39 Won't fit 4.42
Qwen3 8B Runs comfortably 108 Runs comfortably 75.1
Qwen3 14B Runs comfortably 63.5 Runs comfortably 43.9
Qwen3 32B Fits, but tight 29.5 Won't fit 4.34
Qwen3 4B Runs comfortably 196 Runs comfortably 137
Qwen3 30B-A3B Runs comfortably 117 Won't fit 40.2
Qwen3 235B-A22B Won't fit 3.36 Won't fit 3.05
Qwen2.5-Coder 32B Fits, but tight 29.5 Won't fit 4.34
Qwen2.5 7B Runs comfortably 125 Runs comfortably 86.3
Qwen2.5 72B Won't fit 1.59 Won't fit 1.18
Gemma 3 4B Runs comfortably 209 Runs comfortably 146
Gemma 3 12B Runs comfortably 76.1 Runs comfortably 52.7

Specifications

RTX 3090RTX 4070 Ti Super
Memory24 GB16 GB
Bandwidth936 GB/s672 GB/s
FP16 compute71 TFLOPS88 TFLOPS
Launch price$1,499$799
ArchitectureAmpereAda