Nvidia's new free tool, PAIR, lets M4 Macs share local AI workloads with PCs on the same network. It connects Apple Silicon Macs, Nvidia RTX PCs, and DGX Spark systems, working with Ollama and LM Studio to route requests to available machines. PAIR doesn't combine GPUs or memory, so it can't run larger models than a single system could.

In a demo, a task split across five AI subagents ran faster on a DGX Spark and RTX 5090 than on an RTX Spark laptop alone. Mac support requires macOS Tahoe and an M4 or newer chip, so M3 Ultra Macs are not supported. PAIR is in open-source beta as of September 3.