Compatibility check
Can I run Qwen3.5 397B A17B on Apple M3 Ultra (80-core GPU)?
Yes — it runs.best quant: Q4_K_M
Decode speed
47 tok/sest
Usable context
262K
Memory
512 GB
Q4_K_MFits in memory47 tok/sest · 262K ctx
Q5_K_MFits in memory40 tok/sest · 262K ctx
Q8_0Spills to system RAM