can-i-run-this-llm

Can I run this LLM?

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

Kimi · MIT · estimated for a 48.0 GB machine at 4K context.

Yes — it runs great.

Best fit: Kimi Linear 48B-A3B (MoE) at Q5_K_M · 35.1 GB · 172–259 tok/s on a NVIDIA RTX 6000 Ada (48GB).

Every quantization on this machine

QuantSizeSpeedFit
Q2_K17.5 GB233–350 tok/sRuns great
Q3_K_M22.7 GB211–317 tok/sRuns great
Q4_K_M30.1 GB186–280 tok/sRuns great
Q5_K_M35.1 GB172–259 tok/sRuns great
Q6_K40.5 GB160–240 tok/sTight / slow
Q8_052.3 GB137–206 tok/sWon't fit
FP1698.3 GB89–133 tok/sWon't fit
Open the full calculator →Everything a NVIDIA RTX 6000 Ada (48GB) runsKimi Linear 48B-A3B (MoE) on other machinesWhat to buy to run it →

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