Compatibility check
Can I run Granite Speech 3.3 2b on Apple M2 (8-core GPU, 8GB unified)?
Yes — it runs.best quant:
Decode speed
39 tok/sest
Usable context
33K
Memory
8 GB
Fits in memory39 tok/sest · 33K ctx
Fits in memory33 tok/sest · 16K ctx
Fits in memory22 tok/sest · 16K ctx
