GeForce RTX 4090
NVIDIA
24GB VRAM · 1010 GB/s bandwidth · 82.6 FP16 TFLOPS · 450W TDP
The GeForce RTX 4090 has 24GB of VRAM with 1010 GB/s memory bandwidth and 82.6 TFLOPS FP16 compute. At Q4 quantization, it can comfortably run Gemma 3 4B (328 tok/s), Qwen 2.5 7B (172 tok/s), Llama 3.1 8B (163 tok/s). Models larger than ~41B parameters won't fit even at Q4. Electricity cost is approximately $49/month at 450W TDP.
What LLMs can you run?
| Model | Params | Q4 Weight | Fit | Decode |
|---|---|---|---|---|
| Gemma 3 4B | 4.0B | 2 GB | comfortable | 328 tok/s |
| Qwen 2.5 7B | 7.6B | 4 GB | comfortable | 172 tok/s |
| Llama 3.1 8B | 8.0B | 4 GB | comfortable | 163 tok/s |
| Mistral Small 24B | 24.0B | 12 GB | comfortable | 54 tok/s |
| Gemma 3 27B | 27.4B | 14 GB | comfortable | 47 tok/s |
| Qwen 2.5 Coder 32B | 32.5B | 16 GB | comfortable | 40 tok/s |
| Llama 3.3 70B | 70.6B | 35 GB | won't fit | |
| Qwen 2.5 72B | 72.7B | 36 GB | won't fit | |
| Llama 3.1 405B | 405B | 202 GB | won't fit | |
| DeepSeek R1 671B | 671B | 336 GB | won't fit |
Similar GPUs
| GPU | VRAM | BW | TFLOPS | TDP |
|---|---|---|---|---|
| GeForce RTX 3090 Ti | 24GB | 1010 GB/s | 40.0 | 450W |
| GeForce RTX 4090 D | 24GB | 1010 GB/s | 73.5 | 425W |
| Radeon RX 7900 XTX | 24GB | 960 GB/s | 122.8 | 355W |
| GeForce RTX 3090 | 24GB | 936 GB/s | 35.6 | 350W |
| A30 PCIe | 24GB | 933 GB/s | 10.3 | 165W |
Compare with another GPU
Select another GPU to compare specs and model performance side by side.