Can I run this LLM?
Can a NVIDIA DGX Spark (GB10, 128GB) run Kimi Linear 48B-A3B (MoE)?
Kimi · MIT · estimated for a 128.0 GB machine at 4K context.
Yes — it runs great.
Best fit: Kimi Linear 48B-A3B (MoE) at Q8_0 · 52.3 GB · 43–64 tok/s on a NVIDIA DGX Spark (GB10, 128GB).
Every quantization on this machine
| Quant | Size | Speed | Fit |
|---|---|---|---|
| Q2_K | 17.5 GB | 84–127 tok/s | Runs great |
| Q3_K_M | 22.7 GB | 74–111 tok/s | Runs great |
| Q4_K_M | 30.1 GB | 63–94 tok/s | Runs great |
| Q5_K_M | 35.1 GB | 57–85 tok/s | Runs great |
| Q6_K | 40.5 GB | 51–77 tok/s | Runs great |
| Q8_0 | 52.3 GB | 43–64 tok/s | Runs great |
| FP16 | 98.3 GB | 26–39 tok/s | Won't fit |
Open the full calculator →Everything a NVIDIA DGX Spark (GB10, 128GB) runsKimi Linear 48B-A3B (MoE) on other machinesWhat to buy to run it →
Ran Kimi Linear 48B-A3B (MoE) on a NVIDIA DGX Spark (GB10, 128GB)? Add your real numbers →