can it run
Only with compromises. 256GB is below the 307GB working set — expect a lower quant, offloading, or heavy swap. Possible, not pleasant.
weights 245GB + KV cache 0.75GB at 4k context vs ~32GB usable → no. 245.8 GB vs 32 GB usable — does not fit
computed roofline: 1792 GB/s × 0.55–0.80 efficiency window / 245.0GB (Q4_K_M) — bands, never points.
| model file | 245GB (GGUF Q4_K_M) |
| minimum memory | 307GB working set |
| this machine | 256GB VRAM (8× 32GB) |
| params | Maverick 400B A17B |
| license | llama4 (gated on HF) |
| price | $100000 |
easy run: ollama pull llama4:maverick
runs with: ollama · llama.cpp · koboldcpp · lm studio
→ ASUS ExpertCenter Pro ET900N G3 (GB300) — cheapest machine that runs it ($60000)
→ ASUS ExpertCenter Pro ET900N G3 (GB300) — this model on the biggest machine
→ rtx-pro-6000 — the next step up from this machine's GPU