Nvidia's PAIR spreads local AI tasks across home networks
nvidia released pair, a tool that spreads local ai requests across home networks, cutting wait times for parallel tasks. it supports geforce rtx cards and apple silicon.
nvidia introduced pair, a personal ai router that automatically distributes local ai requests across all available devices on a home network, reducing wait times for parallel agent tasks. the tool sits between existing applications like ollama or lm studio and the computers on the network, acting as a virtual router. in a demo, a cluster of three devices completed a task with five subagents in just under 9 minutes, compared to 18 minutes on a single laptop. pair auto-detects compatible devices on the network and secures traffic between machines with mtls encryption. the beta is available for windows, macos, and linux.