can-i-run-this-llm

Can I run this LLM?

Can a NVIDIA H200 (141GB) run Kimi Linear 48B-A3B (MoE)?

Kimi · MIT · estimated for a 141.0 GB machine at 4K context.

Yes — it runs great.

Best fit: Kimi Linear 48B-A3B (MoE) at FP16 · 98.3 GB · 226–339 tok/s on a NVIDIA H200 (141GB).

Every quantization on this machine

QuantSizeSpeedFit
Q2_K17.5 GB329–494 tok/sRuns great
Q3_K_M22.7 GB320–480 tok/sRuns great
Q4_K_M30.1 GB307–461 tok/sRuns great
Q5_K_M35.1 GB299–449 tok/sRuns great
Q6_K40.5 GB291–437 tok/sRuns great
Q8_052.3 GB275–413 tok/sRuns great
FP1698.3 GB226–339 tok/sRuns great

Ran Kimi Linear 48B-A3B (MoE) on a NVIDIA H200 (141GB)? Add your real numbers →