Compatibility check
Can I run GLM 4.5 Air on Apple M3 Ultra (80-core GPU)?
Yes — it runs.best quant:
Decode speed
67 tok/sest
Usable context
131K
Memory
512 GB
Fits in memory67 tok/sest · 131K ctx
Fits in memory57 tok/sest · 131K ctx
Fits in memory38 tok/sest · 131K ctx