can-i-run-this-llm

Can I run this LLM?

Can a AMD Instinct MI300X (192GB) run Llama 3.3 70B Instruct?

Llama · Llama 3.3 Community · estimated for a 192.0 GB machine at 4K context.

Yes — it runs great.

Best fit: Llama 3.3 70B Instruct at FP16 · 141.1 GB · 26–38 tok/s on a AMD Instinct MI300X (192GB).

Every quantization on this machine

QuantSizeSpeedFit
Q2_K26.4 GB102–153 tok/sRuns great
Q3_K_M34.3 GB85–127 tok/sRuns great
Q4_K_M42.5 GB72–108 tok/sRuns great
Q5_K_M50.0 GB63–95 tok/sRuns great
Q6_K57.9 GB56–84 tok/sRuns great
Q8_075.0 GB45–67 tok/sRuns great
FP16141.1 GB26–38 tok/sRuns great

Ran Llama 3.3 70B Instruct on a AMD Instinct MI300X (192GB)? Add your real numbers →