can-i-run-this-llm

Can I run this LLM?

Can a Mac Studio M5 Ultra (256GB) run Hunyuan Hy3 (MoE)?

Hunyuan · Apache 2.0 · estimated for a 256.0 GB machine at 4K context.

Yes — it runs great.

Best fit: Hunyuan Hy3 (MoE) at Q3_K_M · 137.6 GB · 52–78 tok/s on a Mac Studio M5 Ultra (256GB).

Every quantization on this machine

QuantSizeSpeedFit
Q2_K106.7 GB61–91 tok/sRuns great
Q3_K_M137.6 GB52–78 tok/sRuns great
Q4_K_M182.2 GB44–65 tok/sWon't fit
Q5_K_M212.8 GB39–59 tok/sWon't fit
Q6_K257.3 GB34–51 tok/sWon't fit
Q8_0317.7 GB29–44 tok/sWon't fit

Ran Hunyuan Hy3 (MoE) on a Mac Studio M5 Ultra (256GB)? Add your real numbers →