With 80 GB of VRAM, the H100 PCIe runs Qwen3 8B at Q4_K_M entirely on the GPU with an 8K context (about 6.9 GB needed). At Q8_0 it fits entirely.
Q4_K_M · 8K context
The model needs about 6.9 GB and fits in the 77.6 GB available, with 70.7 GB to spare.
~6.9 GB (6.5 GB–7.4 GB) needed · estimate
Fits entirely in VRAM · 70.7 GB left
EstimateData updated 24 days agoSource: Run LLM Locally engine, Hugging Face config.json + model card
What makes this estimate less reliable
| Qwen3 8B Q8_08K~10.5 GB · 122–184 tok/s estimate | Excellent |
|---|---|
| Qwen3 8B Q6_K8K~8.6 GB · 145–226 tok/s estimate | Excellent |
| Qwen3 8B Q5_K_M8K~7.7 GB · 158–251 tok/s estimate | Excellent |
| Qwen3 8B Q4_K_M8K~6.9 GB · 172–280 tok/s estimate | Excellent |
| Qwen3 8B IQ4_XS8K~6.5 GB · 182–300 tok/s estimate | Excellent |
| Qwen3 8B Q3_K_M8K~6.1 GB · 192–321 tok/s estimate | Excellent |
Prices and links follow the market you choose.
NVIDIA
CompatibleRuns Qwen3 8B Q4_K_M entirely in memory
AMD
ExcellentRuns Qwen3 8B Q4_K_M entirely in memory
Unified memory PC
ExcellentRuns Qwen3 8B Q4_K_M entirely in memory
No current offer imported for this market: the link opens a marketplace search.
