Compatibility check
Can I run Llama 3 8B Lunaris on Apple M1 Ultra (64-core GPU)?
Yes — it runs.best quant:
Decode speed
116 tok/sest
Usable context
8K
Memory
128 GB
Fits in memory116 tok/sest · 8K ctx
Fits in memory98 tok/sest · 8K ctx
Fits in memory66 tok/sest · 8K ctx