can-i-run-this-llm

Can I run this LLM?

Can a NVIDIA H100 (80GB) run Kimi Linear 48B-A3B (MoE)?

Kimi · MIT · estimated for a 80.0 GB machine at 4K context.

Yes — it runs great.

Best fit: Kimi Linear 48B-A3B (MoE) at Q8_0 · 52.2 GB · 248–371 tok/s on a NVIDIA H100 (80GB).

Every quantization on this machine

QuantSizeSpeedFit
Q2_K16.0 GB318–477 tok/sRuns great
Q3_K_M21.2 GB305–458 tok/sRuns great
Q4_K_M29.8 GB287–430 tok/sRuns great
Q5_K_M35.0 GB277–415 tok/sRuns great
Q6_K40.5 GB267–400 tok/sRuns great
Q8_052.2 GB248–371 tok/sRuns great
FP1698.2 GB193–290 tok/sWon't fit

Ran Kimi Linear 48B-A3B (MoE) on a NVIDIA H100 (80GB)? Add your real numbers →