Compatibility check
Can I run Lite Whisper Large v3 Acc on Apple M4 Max (40-core GPU)?
Yes — it runs.best quant:
Decode speed
450 tok/sest
Usable context
262K
Memory
128 GB
Fits in memory450 tok/sest · 262K ctx
Fits in memory384 tok/sest · 262K ctx
Fits in memory257 tok/sest · 262K ctx