local intelligence

can it run

Can the Apple M2 Ultra run LLAMA4 Scout 109B A17B?

OFFLOAD PARTIAL

67.8 GB vs 62 GB usable — partial CPU offload, expect large speed loss With a q8 (quantized) KV cache it comes to the edge (67.4GB) — possible with modest context, at some quality cost.

~62GB usable · Apple M2 Ultra · model file 67GB · Scout 109B A17B

~6.6–9.6 tok/s

computed roofline: 800 GB/s × 0.55–0.80 efficiency window / 67.0GB (Q4_K_M) — bands, never points.

Run LLAMA4 Scout 109B A17B on instead

NIMO AI Mini PC (128GB) — cheapest catalog machine that runs LLAMA4 Scout 109B A17B fully ($1999)

every model the Apple M2 Ultra can run — full list

explore the full catalog84 machines indexed · live prices · what each one can run
2026-09-17 · ← all models on the Apple M2 Ultra · fit = weights + 4k KV vs usable memory