local intelligence

can it run

Can the Radeon PRO W7800 32GB run qwen3 235B-A22B MoE?

NO

142.7 GB vs 32 GB usable — does not fit

32GB · RDNA3 · model file 142GB · 235B-A22B MoE

~22–32 tok/s (active-expert estimate)

computed roofline: 576 GB/s × 0.50–0.75 efficiency window / 13.3GB (Q4_K_M) — bands, never points.

MoE: file is 142GB (that decides fit) but only ~22B of experts activate per token (~13.3GB streamed) — band below is the active-expert estimate, optimistic; community MoE rows run several× the naive bandwidth math

Run qwen3 235B-A22B MoE on instead

ASUS ExpertCenter Pro ET900N G3 (GB300) — cheapest catalog machine that runs qwen3 235B-A22B MoE fully ($60000)

Radeon AI PRO R9700 (32GB) — upgrade path machine

every model the Radeon PRO W7800 32GB can run — full list

explore the full catalog90 machines indexed · live prices · what each one can run
2026-09-24 · ← all models on the Radeon PRO W7800 32GB · fit = weights + 4k KV vs usable memory