Nvidia’s PAIR turns home networks into local AI clusters
Nvidia has introduced PAIR, or Personal AI Router, an open-source tool designed to distribute local AI workloads across devices connected to a home network. It operates between applications such as Ollama or LM Studio and the computers running them, so users do not need to modify their agents or apps.
PAIR identifies compatible machines automatically and sends requests to available systems instead of concentrating them on one GPU. It then combines the outputs for the application that initiated the task. Nvidia says supported hardware includes GeForce RTX cards from the 20 series onward, RTX Pro workstations, DGX Spark, and Apple silicon beginning with the M4. Traffic between devices is protected with MTLS encryption.
In a demonstration, three devices completed a task involving five subagents in just under nine minutes, compared with 18 minutes on one laptop. The beta is available for Windows, macOS, and Linux. The tool is part of Nvidia’s wider effort to connect open AI software with its hardware, also reflected in its $12.9 billion acquisition of Hugging Face.
Written by Sakhanda Wire AI from reporting by The Decoder. Facts, quotes and full context are in the original article below.