With 48 GB of VRAM, the RTX 6000 Ada runs Llama 3.2 3B Instruct at Q4_K_M entirely on the GPU with an 8K context (about 3.8 GB needed). At Q8_0 it fits entirely.
Q4_K_M · 8K context
The model needs about 3.8 GB and fits in the 46.6 GB available, with 42.8 GB to spare.
~3.8 GB (3.5 GB–4.2 GB) needed · estimate
Fits entirely in VRAM · 42.8 GB left
EstimateData updated 24 days agoSource: Run LLM Locally engine, Hugging Face config.json + model card
What makes this estimate less reliable
| Llama 3.2 3B Instruct Q8_08K~5.1 GB · 144–214 tok/s estimate | Excellent |
|---|---|
| Llama 3.2 3B Instruct Q6_K8K~4.4 GB · 170–260 tok/s estimate | Excellent |
| Llama 3.2 3B Instruct Q5_K_M8K~4.1 GB · 184–285 tok/s estimate | Excellent |
| Llama 3.2 3B Instruct Q4_K_M8K~3.8 GB · 199–314 tok/s estimate | Excellent |
| Llama 3.2 3B Instruct IQ4_XS8K~3.6 GB · 209–335 tok/s estimate | Excellent |
| Llama 3.2 3B Instruct Q3_K_M8K~3.5 GB · 218–353 tok/s estimate | Excellent |
Prices and links follow the market you choose.
48 GB GDDR6 ECC · 960 GB/s · 300 W
ExcellentRuns Llama 3.2 3B Instruct Q4_K_M entirely in memory
No current offer imported for this market: the link opens a marketplace search.
