Head to head

RTX 3090 vs RTX 5070 Ti

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 5070 Titok/s
Llama 3.1 8B Runs comfortably 112 Runs comfortably 113
Llama 3.1 70B Won't fit 1.67 Won't fit 1.35
Llama 3.2 3B Runs comfortably 246 Runs comfortably 249
Llama 3.2 1B Runs comfortably 650 Runs comfortably 657
Llama 4 Scout 109B-A17B Won't fit 5.39 Won't fit 4.90
Qwen3 8B Runs comfortably 108 Runs comfortably 109
Qwen3 14B Runs comfortably 63.5 Runs comfortably 64.2
Qwen3 32B Fits, but tight 29.5 Won't fit 4.95
Qwen3 4B Runs comfortably 196 Runs comfortably 198
Qwen3 30B-A3B Runs comfortably 117 Won't fit 45.7
Qwen3 235B-A22B Won't fit 3.36 Won't fit 3.36
Qwen2.5-Coder 32B Fits, but tight 29.5 Won't fit 4.95
Qwen2.5 7B Runs comfortably 125 Runs comfortably 126
Qwen2.5 72B Won't fit 1.59 Won't fit 1.31
Gemma 3 4B Runs comfortably 209 Runs comfortably 211
Gemma 3 12B Runs comfortably 76.1 Runs comfortably 76.9

Specifications

RTX 3090RTX 5070 Ti
Memory24 GB16 GB
Bandwidth936 GB/s896 GB/s
FP16 compute71 TFLOPS88 TFLOPS
Launch price$1,499$749
ArchitectureAmpereBlackwell