local intelligence

can it run

Can the Intel Arc A580 run qwen3 8B?

FULL

5.8 GB needed of 8 GB usable — headroom for context

8GB · Alchemist · model file 5.2GB · 8B

~20–49 tok/s

computed roofline: 512 GB/s × 0.20–0.50 efficiency window / 5.2GB (Q4_K_M) — bands, never points.

Run qwen3 8B on instead

Tesla P100 (16GB, used) — cheapest catalog machine that runs qwen3 8B fully ($135)

arc-a770-16gb — the next step up from Intel Arc A580

every model the Intel Arc A580 can run — full list

explore the full catalog84 machines indexed · live prices · what each one can run
2026-09-17 · ← all models on the Intel Arc A580 · fit = weights + 4k KV vs usable memory