can-i-run-this-llm

Can I run this LLM?

Can a Mac Studio M5 Ultra (512GB) run Llama 3.3 70B Instruct?

Llama · Llama 3.3 Community · estimated for a 512.0 GB machine at 4K context.

Yes — it runs great.

Best fit: Llama 3.3 70B Instruct at Q6_K · 57.9 GB · 13–19 tok/s on a Mac Studio M5 Ultra (512GB).

Every quantization on this machine

QuantSizeSpeedFit
Q2_K26.4 GB25–38 tok/sRuns great
Q3_K_M34.3 GB20–31 tok/sRuns great
Q4_K_M42.5 GB17–26 tok/sRuns great
Q5_K_M50.0 GB15–22 tok/sRuns great
Q6_K57.9 GB13–19 tok/sRuns great
Q8_075.0 GB10–15 tok/sTight / slow
FP16141.1 GB6–8 tok/sTight / slow
Open the full calculator →Everything a Mac Studio M5 Ultra (512GB) runsLlama 3.3 70B Instruct on other machinesWhat to buy to run it →

Ran Llama 3.3 70B Instruct on a Mac Studio M5 Ultra (512GB)? Add your real numbers →