local intelligence

can it run

Can the AMD Instinct MI300A 128GB run qwen3 1.7B?

FULL

1.8 GB needed of 128 GB usable — headroom for context

128GB · CDNA3 · model file 1.4GB · 1.7B

~1100–1700 tok/s

computed roofline: 5300 GB/s × 0.28–0.45 efficiency window / 1.4GB (Q4_K_M) — bands, never points. Small models (under ~4.5B) run BELOW these bands — kernel-launch overhead keeps them from saturating bandwidth (measured against community medians, 09-06).

Run qwen3 1.7B on instead

Tesla M60 (16GB, used) — cheapest catalog machine that runs qwen3 1.7B fully ($35)

NVIDIA DGX Station (GB300) — upgrade path machine

every model the AMD Instinct MI300A 128GB can run — full list

explore the full catalog90 machines indexed · live prices · what each one can run
2026-09-24 · ← all models on the AMD Instinct MI300A 128GB · fit = weights + 4k KV vs usable memory