Kimi K3 Q2 reportedly runs locally on two 512GB M3 Ultra Mac Studios
pcuenq · x · 2026-07-29
Kimi K3 Q2 is reportedly running locally on two Mac Studio machines with M3 Ultra 512GB each. The quote adds that the setup uses pi agent plus mlx-lm, suggesting a practical local-deployment path for a very large model.
- Main point: a 2.8T-parameter Kimi model is being shown running on consumer Mac hardware.
- Stack mentioned: two Mac Studios, pi agent, and mlx-lm.
- Why it matters: it highlights the latest boundary of local inference and model portability.
More from Infra
- Meta's commitments jumped to $349B in a quarter as server memory prices tripled ahead of Oct 21 earnings — tengyanAI · 2026-10-16
- Meta releases Muse Spark Open, trained with vLLM-powered RL inference from day one — colinmcnamara · 2026-10-16
- Meta's open Muse Spark needs 493GB of memory — far beyond any consumer laptop — TejasKumar_ · 2026-10-16
- CoreWeave plunges 20% as credit markets price >60% default odds; is this dot-com all over again? — kimmonismus · 2026-10-16
- MiniCPM-V 4.7 35B-A3B hits 17+ tokens/s on iPad with ~3GB memory — techNmak · 2026-10-16
- China rolls out 'token factories' and a 2026-28 token economy action plan — nordicinst · 2026-10-16