RTX 4000 SFF Ada Generation
NVIDIA
20GB VRAM · 280 GB/s bandwidth · 19.2 FP16 TFLOPS · 70W TDP
The RTX 4000 SFF Ada Generation has 20GB of VRAM with 280 GB/s memory bandwidth and 19.2 TFLOPS FP16 compute. At Q4 quantization, it can comfortably run Gemma 3 4B (91 tok/s), Qwen 2.5 7B (47 tok/s), Llama 3.1 8B (45 tok/s). Models larger than ~34B parameters won't fit even at Q4. Electricity cost is approximately $8/month at 70W TDP.
What LLMs can you run?
| Model | Params | Q4 Weight | Fit | Decode |
|---|---|---|---|---|
| Gemma 3 4B | 4.0B | 2 GB | comfortable | 91 tok/s |
| Qwen 2.5 7B | 7.6B | 4 GB | comfortable | 47 tok/s |
| Llama 3.1 8B | 8.0B | 4 GB | comfortable | 45 tok/s |
| Mistral Small 24B | 24.0B | 12 GB | tight | 15 tok/s |
| Gemma 3 27B | 27.4B | 14 GB | tight | 13 tok/s |
| Qwen 2.5 Coder 32B | 32.5B | 16 GB | tight | 11 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 |
|---|---|---|---|---|
| RTX 4000 Ada Generation | 20GB | 360 GB/s | 26.7 | 130W |
| A10M | 20GB | 500 GB/s | 23.4 | 150W |
| RTX A4500 | 20GB | 640 GB/s | 23.6 | 200W |
| GeForce RTX 3080 Ti 20 GB | 20GB | 760 GB/s | 34.1 | 350W |
| Radeon RX 7900 XT | 20GB | 800 GB/s | 103.0 | 300W |
Compare with another GPU
Select another GPU to compare specs and model performance side by side.