Compatibility check
Can I run Qwen3 ASR 1.7B HF on Apple M3 Ultra (80-core GPU)?
Yes — it runs.best quant:
Decode speed
473 tok/sest
Usable context
262K
Memory
512 GB
Fits in memory473 tok/sest · 262K ctx
Fits in memory403 tok/sest · 262K ctx
Fits in memory270 tok/sest · 262K ctx