Can I run this LLM?
Can a Mac Studio M3 Ultra (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 · 86–129 tok/s on a Mac Studio M3 Ultra (96GB).
Every quantization on this machine
| Quant | Size | Speed | Fit |
|---|---|---|---|
| Q2_K | 17.5 GB | 128–192 tok/s | Runs great |
| Q3_K_M | 22.7 GB | 119–179 tok/s | Runs great |
| Q4_K_M | 30.1 GB | 109–163 tok/s | Runs great |
| Q5_K_M | 35.1 GB | 102–154 tok/s | Runs great |
| Q6_K | 40.5 GB | 96–145 tok/s | Runs great |
| Q8_0 | 52.3 GB | 86–129 tok/s | Runs great |
| FP16 | 98.3 GB | 60–90 tok/s | Won't fit |
Open the full calculator →Everything a Mac Studio M3 Ultra (96GB) runsKimi Linear 48B-A3B (MoE) on other machinesWhat to buy to run it →
Ran Kimi Linear 48B-A3B (MoE) on a Mac Studio M3 Ultra (96GB)? Add your real numbers →