hey have u guys tried NVIDIA PAIR? it just dropped at IFA 2026 and it's actually pretty smart
it's a free open source tool that routes AI inference across all the PCs on your local network. so if u have a desktop with an RTX GPU and a laptop just sitting there idle it sends the next request to whichever machine has free compute
works with Ollama and LM Studio. no code changes needed to your agents
the catch is it doesn't pool VRAM or shard models across machines. it just picks the best free node for each request. so it's a router not a cluster
thinking about setting it up with my home setup. if anyone's already running it lmk how it goes
Top comments (0)