can it run
Comfortably. 384GB of memory against a 54GB working set leaves headroom for context, the OS, and anything else you're running.
weights 43GB + KV cache 1.25GB at 4k context vs ~96GB usable → full. 44.2 GB needed of 96 GB usable — headroom for context
computed roofline: 1792 GB/s × 0.46–0.76 efficiency window / 43.0GB (Q4_K_M) — bands, never points.
| model file | 43GB (GGUF Q4_K_M) |
| minimum memory | 54GB working set |
| this machine | 384GB VRAM (4× 96GB) |
| params | 70B |
| license | llama3.1 (gated on HF) |
| price | $91000 |
easy run: ollama pull llama3.1:70b
runs with: ollama · llama.cpp · koboldcpp · lm studio
→ ASUS ExpertCenter Pro ET900N G3 (GB300) — this model on the biggest machine