Head to head

RTX 4090 vs RTX 3060 12GB

The RTX 4090 holds more — 24 GB against 12 GB — which decides what you can load at all. The RTX 4090 has more bandwidth at 1008 GB/s, which decides how fast tokens come out once a model fits.

Model (Q4_K_M, 8K)RTX 4090tok/sRTX 3060 12GBtok/s
Llama 3.1 8B Runs comfortably 116 Runs comfortably 43.7
Llama 3.1 70B Won't fit 1.60 Won't fit 1.13
Llama 3.2 3B Runs comfortably 254 Runs comfortably 97.2
Llama 3.2 1B Runs comfortably 669 Runs comfortably 260
Llama 4 Scout 109B-A17B Won't fit 5.18 Won't fit 4.27
Qwen3 8B Runs comfortably 112 Runs comfortably 42.3
Qwen3 14B Runs comfortably 65.5 Won't fit 22.1
Qwen3 32B Fits, but tight 30.5 Won't fit 3.06
Qwen3 4B Runs comfortably 202 Runs comfortably 77.4
Qwen3 30B-A3B Runs comfortably 118 Won't fit 26.1
Qwen3 235B-A22B Won't fit 3.23 Won't fit 3.07
Qwen2.5-Coder 32B Fits, but tight 30.5 Won't fit 3.06
Qwen2.5 7B Runs comfortably 128 Runs comfortably 48.5
Qwen2.5 72B Won't fit 1.53 Won't fit 1.08
Gemma 3 4B Runs comfortably 215 Runs comfortably 82.6
Gemma 3 12B Runs comfortably 78.5 Runs comfortably 29.7

Specifications

RTX 4090RTX 3060 12GB
Memory24 GB12 GB
Bandwidth1008 GB/s360 GB/s
FP16 compute165 TFLOPS25 TFLOPS
Launch price$1,599$329
ArchitectureAdaAmpere