With 12 GB of VRAM, the RTX 4070 runs Granite 4.2 8B at Q4_K_M entirely on the GPU with an 8K context (about 7.3 GB needed). At Q8_0 it fits entirely.
Q4_K_M · 8K context
The model needs about 7.3 GB and fits in the 11.6 GB available, with 4.4 GB to spare.
~7.3 GB (6.9 GB–7.7 GB) needed · estimate
Fits entirely in VRAM · 4.4 GB left
EstimateData updated 40 days agoSource: Run LLM Locally engine, Hugging Face config.json + model card
What makes this estimate less reliable
| Granite 4.2 8B Q8_08K~11.2 GB · 36–48 tok/s estimate | Compatible with limits |
|---|---|
| Granite 4.2 8B Q6_K8K~9.1 GB · 44–60 tok/s estimate | Excellent |
| Granite 4.2 8B Q5_K_M8K~8.2 GB · 50–68 tok/s estimate | Excellent |
| Granite 4.2 8B Q4_K_M8K~7.3 GB · 56–78 tok/s estimate | Excellent |
| Granite 4.2 8B IQ4_XS8K~6.8 GB · 60–83 tok/s estimate | Excellent |
| Granite 4.2 8B Q3_K_M8K~6.3 GB · 65–92 tok/s estimate | Excellent |
Prices and links follow the market you choose.
12 GB GDDR6X · 504 GB/s · 200 W
ExcellentRuns Granite 4.2 8B Q4_K_M entirely in memory
No current offer imported for this market: the link opens a marketplace search.
