Compatibility check
Can I run MOSS Transcribe Preview 2B on Apple M5 Max (40-core GPU)?
Yes — it runs.best quant:
Decode speed
295 tok/sest
Usable context
33K
Memory
128 GB
Fits in memory295 tok/sest · 33K ctx
Fits in memory252 tok/sest · 33K ctx
Fits in memory169 tok/sest · 33K ctx