With 48 GB of VRAM, the RTX 6000 Ada runs DeepSeek R1 Distill Llama 70B at Q4_K_M entirely on the GPU with an 8K context (about 45.1 GB needed). At Q8_0 it does not fit in VRAM.
Q4_K_M · 8K context
The model needs about 45.1 GB and fits in the 46.6 GB available (1.5 GB left).
~45.1 GB (43.9 GB–46.3 GB) needed · estimate
Fits entirely in VRAM · 1.5 GB left
EstimateData updated 24 days agoSource: Run LLM Locally engine, Hugging Face config.json + model card
What makes this estimate less reliable
| DeepSeek R1 Distill Llama 70B Q8_08K~76.8 GB · 0.6–1.2 tok/s estimate | Not recommended |
|---|---|
| DeepSeek R1 Distill Llama 70B Q6_K8K~60.1 GB · 1.3–2.5 tok/s estimate | Possible but slow |
| DeepSeek R1 Distill Llama 70B Q5_K_M8K~52.3 GB · 2.5–5 tok/s estimate | CPU/GPU offload |
| DeepSeek R1 Distill Llama 70B Q4_K_M8K~45.1 GB · 16–21 tok/s estimate | Compatible |
| DeepSeek R1 Distill Llama 70B IQ4_XS8K~40.6 GB · 18–23 tok/s estimate | Compatible |
| DeepSeek R1 Distill Llama 70B Q3_K_M8K~37 GB · 19–26 tok/s estimate | Excellent |
Prices and links follow the market you choose.
48 GB GDDR6 ECC · 960 GB/s · 300 W
CompatibleRuns DeepSeek R1 Distill Llama 70B Q4_K_M entirely in memory
No current offer imported for this market: the link opens a marketplace search.
