can-i-run-this-llm

Can I run this LLM?

Can a NVIDIA RTX Pro 6000 Blackwell (96GB) run Kimi Linear 48B-A3B (MoE)?

Kimi · MIT · estimated for a 96.0 GB machine at 4K context.

Yes — it runs great.

Best fit: Kimi Linear 48B-A3B (MoE) at Q8_0 · 52.3 GB · 194–291 tok/s on a NVIDIA RTX Pro 6000 Blackwell (96GB).

Every quantization on this machine

QuantSizeSpeedFit
Q2_K17.5 GB281–421 tok/sRuns great
Q3_K_M22.7 GB263–395 tok/sRuns great
Q4_K_M30.1 GB242–362 tok/sRuns great
Q5_K_M35.1 GB229–343 tok/sRuns great
Q6_K40.5 GB216–325 tok/sRuns great
Q8_052.3 GB194–291 tok/sRuns great
FP1698.3 GB137–206 tok/sWon't fit

Ran Kimi Linear 48B-A3B (MoE) on a NVIDIA RTX Pro 6000 Blackwell (96GB)? Add your real numbers →