Compatibility check
Can I run MOSS Transcribe Preview 2B on Apple M3 Ultra (80-core GPU)?
Yes — it runs.best quant:
Decode speed
394 tok/sest
Usable context
33K
Memory
512 GB
Fits in memory394 tok/sest · 33K ctx
Fits in memory336 tok/sest · 33K ctx
Fits in memory225 tok/sest · 33K ctx
