can-i-run-this-llm

Can I run this LLM?

Can a AMD Ryzen AI Max+ 395 “Strix Halo” (64GB) run Gemma 4 31B?

Gemma · Apache 2.0 · estimated for a 64.0 GB machine at 4K context.

Yes, but it's tight.

Best fit: Gemma 4 31B at Q3_K_M · 14.7 GB · 10–15 tok/s on a AMD Ryzen AI Max+ 395 “Strix Halo” (64GB).

Every quantization on this machine

QuantSizeSpeedFit
Q3_K_M14.7 GB10–15 tok/sTight / slow
Q4_K_M18.3 GB8–13 tok/sTight / slow
Q5_K_M21.7 GB7–11 tok/sTight / slow
Q6_K25.2 GB6–9 tok/sTight / slow
Q8_032.6 GB5–7 tok/sTight / slow
FP1661.4 GB3–4 tok/sWon't fit

Ran Gemma 4 31B on a AMD Ryzen AI Max+ 395 “Strix Halo” (64GB)? Add your real numbers →