
NVIDIA PAIR turns idle home PCs into a local AI cluster
NVIDIA has released PAIR, a free, open-source software tool that aggregates idle computing power from multiple home devices to run local AI inference. By routing tasks across available GPUs on a local network, PAIR aims to reduce reliance on cloud APIs and improve performance for multi-agent workflows. The beta version supports Windows, Linux, and macOS, compatible with NVIDIA RTX 20-series and newer GPUs, DGX Spark systems, and Apple M4 silicon. It integrates with existing tools like Ollama and LM Studio, requiring no new hardware or API changes.


