With 16 GB of VRAM, the RX 6800 XT runs Llama 3.2 3B Instruct at Q4_K_M entirely on the GPU with an 8K context (about 3.7 GB needed). At Q8_0 it fits entirely.
Q4_K_M · 8K context
The model needs about 3.7 GB and fits in the 15.5 GB available, with 11.8 GB to spare.
~3.7 GB (3.4 GB–4.1 GB) needed · estimate
Fits entirely in VRAM · 11.8 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 · 74–120 tok/s estimate | Excellent |
|---|---|
| Llama 3.2 3B Instruct Q6_K8K~4.3 GB · 90–149 tok/s estimate | Excellent |
| Llama 3.2 3B Instruct Q5_K_M8K~4 GB · 99–165 tok/s estimate | Excellent |
| Llama 3.2 3B Instruct Q4_K_M8K~3.7 GB · 108–184 tok/s estimate | Excellent |
| Llama 3.2 3B Instruct IQ4_XS8K~3.6 GB · 116–199 tok/s estimate | Excellent |
| Llama 3.2 3B Instruct Q3_K_M8K~3.4 GB · 122–211 tok/s estimate | Excellent |
Prices and links follow the market you choose.
16 GB GDDR6 · 512 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.
