Models /Llama 3.1 70B Instruct /Can I run it?

Compatibility check

Can I run Llama 3.1 70B Instruct on Apple M2 Max (38-core GPU)?

Yes — it runs.best quant:

Decode speed

7 tok/sest

Usable context

131K

Memory

96 GB

Fits in memory7 tok/sest · 131K ctx

Fits in memory6 tok/sest · 66K ctx

Spills to system RAM

What is quantisation? →

Llama 3.1 70B Instruct

70.6B · full specs & benchmarks →

Apple M2 Max (38-core GPU)

96 GB · everything it runs →