With 48 GB of VRAM, the RTX 6000 Ada runs Qwen2.5 7B Instruct at Q4_K_M entirely on the GPU with an 8K context (about 5.9 GB needed). At Q8_0 it fits entirely.
Q4_K_M · 8K context
The model needs about 5.9 GB and fits in the 46.6 GB available, with 40.7 GB to spare.
~5.9 GB (5.5 GB–6.3 GB) needed · estimate
Fits entirely in VRAM · 40.7 GB left
EstimateData updated 24 days agoSource: Run LLM Locally engine, Hugging Face config.json + model card
What makes this estimate less reliable
| Qwen2.5 7B Instruct Q8_08K~9.2 GB · 77–107 tok/s estimate | Excellent |
|---|---|
| Qwen2.5 7B Instruct Q6_K8K~7.4 GB · 96–135 tok/s estimate | Excellent |
| Qwen2.5 7B Instruct Q5_K_M8K~6.6 GB · 107–153 tok/s estimate | Excellent |
| Qwen2.5 7B Instruct Q4_K_M8K~5.9 GB · 120–174 tok/s estimate | Excellent |
| Qwen2.5 7B Instruct IQ4_XS8K~5.5 GB · 130–190 tok/s estimate | Excellent |
| Qwen2.5 7B Instruct Q3_K_M8K~5.1 GB · 140–207 tok/s estimate | Excellent |
Prices and links follow the market you choose.
48 GB GDDR6 ECC · 960 GB/s · 300 W
ExcellentRuns Qwen2.5 7B Instruct Q4_K_M entirely in memory
No current offer imported for this market: the link opens a marketplace search.
