can it run
No. 48GB of memory can't hold the 205GB working set this model needs. A smaller quant or a bigger machine is required.
weights 164GB + KV cache 2GB at 4k context vs ~48GB usable → no. 166.0 GB vs 48 GB usable — does not fit
computed roofline: 864 GB/s × 0.46–0.76 efficiency window / 9.2GB (Q4_K_M) — bands, never points.
MoE: file is 164GB (that decides fit) but only ~18B of experts activate per token (~9.2GB streamed) — band below is the active-expert estimate, optimistic; community MoE rows run several× the naive bandwidth math
| model file | 164GB (GGUF Q4_K_M) |
| minimum memory | 205GB working set |
| this machine | 48GB GDDR6, 864 GB/s |
| params | 320B-A18B MoE |
| license | MIT |
| price | $7450 |
easy run: no stock runtime yet
→ ASUS ExpertCenter Pro ET900N G3 (GB300) — cheapest machine that runs it ($60000)
→ rtx-pro-6000 — the next step up from this machine's GPU
NVIDIA (used market) ↗ · Amazon ↗