can-i-run-this-llm

Can I run this LLM?

Can a Mac Studio M1 Ultra (128GB) run Kimi Linear 48B-A3B (MoE)?

Kimi · MIT · estimated for a 128.0 GB machine at 4K context.

Yes — it runs great.

Best fit: Kimi Linear 48B-A3B (MoE) at Q8_0 · 52.3 GB · 85–127 tok/s on a Mac Studio M1 Ultra (128GB).

Every quantization on this machine

QuantSizeSpeedFit
Q2_K17.5 GB127–190 tok/sRuns great
Q3_K_M22.7 GB118–177 tok/sRuns great
Q4_K_M30.1 GB108–161 tok/sRuns great
Q5_K_M35.1 GB101–152 tok/sRuns great
Q6_K40.5 GB95–143 tok/sRuns great
Q8_052.3 GB85–127 tok/sRuns great
FP1698.3 GB59–88 tok/sWon't fit

Ran Kimi Linear 48B-A3B (MoE) on a Mac Studio M1 Ultra (128GB)? Add your real numbers →