can it run
Comfortably. 256GB of memory against a 11GB working set leaves headroom for context, the OS, and anything else you're running.
weights 8.1GB + KV cache 1.5GB at 4k context vs ~32GB usable → full. 9.6 GB needed of 32 GB usable — headroom for context
computed roofline: 1792 GB/s × 0.55–0.80 efficiency window / 8.1GB (Q4_K_M) — bands, never points.
| model file | 8.1GB (GGUF Q4_K_M) |
| minimum memory | 11GB working set |
| this machine | 256GB VRAM (8× 32GB) |
| params | 12B |
| license | gemma |
| price | $100000 |
easy run: ollama pull gemma3:12b
runs with: ollama · llama.cpp · koboldcpp · lm studio
→ ASUS ExpertCenter Pro ET900N G3 (GB300) — this model on the biggest machine
→ rtx-pro-6000 — the next step up from this machine's GPU