Heterogeneous RTX 6000 + M5 Laptop Cluster Runs Local LLM at 40 tok/s

Hugging Face engineer pcuenq demonstrated distributed inference across heterogeneous hardware, running MiMo 2.6 Flash on an RTX 6000 Blackwell plus M5 laptop over 10 GbE with native mxfp4 weights, achieving 40 tokens/sec in a DIY mini cluster.

2026-10-07 ~ 2026-10-07 · 3 related posts

• Running MiMo 2.6 Flash across an RTX 6000 and M5 laptop at 40 tokens/sec over 10 GbE — pcuenq · 2026-10-07
• Running MiMo 2.6 Flash across an RTX 6000 and M5 laptop at 40 tokens/sec over 10 GbE — pcuenq · 2026-10-07
• A DIY micro-cluster: M5 Max 128GB plus Linux box with RTX 6000 for local AI — pcuenq · 2026-10-07