Question 1

Can the Radeon PRO W7800 32GB run GLM-5.3-Flash (Ox Alpha)?

Accepted Answer

GLM-5.3-Flash (Ox Alpha) needs about 166GB working set (weights 164GB + KV cache 2GB at 4k context). The Radeon PRO W7800 32GB offers ~32GB usable — verdict: no. 166.0 GB vs 32 GB usable — does not fit

Question 2

How fast is GLM-5.3-Flash (Ox Alpha) on the Radeon PRO W7800 32GB?

Accepted Answer

~31–47 tok/s (active-expert estimate) on the Radeon PRO W7800 32GB. computed roofline: 576 GB/s × 0.50–0.75 efficiency window / 9.2GB (Q4_K_M) — bands, never points. MoE: file is 164GB (that decides fit) but only ~18B of experts activate per token (~9.2GB streamed) — band below is the active-expert estimate, optimistic; community MoE rows run several× the naive bandwidth math

Question 3

What should I run GLM-5.3-Flash (Ox Alpha) on instead?

Accepted Answer

ASUS ExpertCenter Pro ET900N G3 (GB300) runs it fully ($60000).

Can the Radeon PRO W7800 32GB run GLM-5.3-Flash (Ox Alpha)?

Run GLM-5.3-Flash (Ox Alpha) on instead