can it run
Comfortably. 128GB of memory against a 4GB working set leaves headroom for context, the OS, and anything else you're running.
weights 0.3GB + KV cache 2GB at 4k context vs ~126GB usable → full. 2.3 GB needed of 126 GB usable — headroom for context
computed roofline: 273 GB/s × 0.25–0.65 efficiency window / 0.3GB (Q4_K_M) — bands, never points.
| model file | 0.3GB (GGUF Q4_K_M) |
| minimum memory | 4GB working set |
| this machine | 128GB unified LPDDR5x |
| params | 82M |
| license | apache-2.0 |
| price | $4699 |
easy run: no stock runtime yet
→ every model the NVIDIA DGX Spark (GB10) can run — full list
→ rtx-5090 — the next step up from this machine's GPU
NVIDIA (direct) ↗ · Micro Center ↗ · Best Buy ↗