vram.run Models Hardware Providers Cloud State of Inference
API provider data is live · Hardware & cloud pricing curated 2026-02-23

RTX A6000

NVIDIA

48GB VRAM · 768 GB/s bandwidth · 38.7 FP16 TFLOPS · 300W TDP

The RTX A6000 has 48GB of VRAM with 768 GB/s memory bandwidth and 38.7 TFLOPS FP16 compute. At Q4 quantization, it can comfortably run Gemma 3 4B (238 tok/s), Qwen 2.5 7B (125 tok/s), Llama 3.1 8B (118 tok/s). Models larger than ~82B parameters won't fit even at Q4. Electricity cost is approximately $32/month at 300W TDP.

What LLMs can you run?

ModelParamsQ4 WeightFitDecode
Gemma 3 4B4.0B2 GBcomfortable238 tok/s
Qwen 2.5 7B7.6B4 GBcomfortable125 tok/s
Llama 3.1 8B8.0B4 GBcomfortable118 tok/s
Mistral Small 24B24.0B12 GBcomfortable39 tok/s
Gemma 3 27B27.4B14 GBcomfortable34 tok/s
Qwen 2.5 Coder 32B32.5B16 GBtight29 tok/s
Llama 3.3 70B70.6B35 GBtight13 tok/s
Qwen 2.5 72B72.7B36 GBtight13 tok/s
Llama 3.1 405B405B202 GBwon't fit
DeepSeek R1 671B671B336 GBwon't fit

Similar GPUs

GPUVRAMBWTFLOPSTDP
A40 PCIe48GB695 GB/s37.4300W
Radeon PRO W7800 48 GB48GB864 GB/s90.5281W
Radeon PRO W790048GB864 GB/s122.6295W
Radeon PRO W7900D48GB864 GB/s106.0295W
Quadro RTX 800048GB672 GB/s32.6260W

Compare with another GPU

Select another GPU to compare specs and model performance side by side.

Install CLI [email protected] Raw data · MIT · API data: live · HW/Cloud data: curated 2026-02-23 · v0.6.0