NVIDIA Releases Personal AI Router (PAIR): An Open Source Virtual Inference Router that Distributes Local AI Requests Across RTX, DGX Spark, and Mac Nodes
NVIDIA's PAIR lets you route local AI inference across your home cluster—no agent code changes needed.
Why it matters
PAIR is open-source software that enables practitioners running multiple AI devices (RTX, DGX Spark, Macs) to pool compute and balance inference loads transparently via Ollama/LM Studio proxying. It lowers the friction for home/lab-scale multi-device AI deployments and agent orchestration.
The key facts
12 to knowOpen source virtual inference router for local/home networks
Proxies existing Ollama and LM Studio endpoints without agent code changes
Scheduler filters on node readiness, engine state, model presence, job load, GPU utilization
Demonstration: 5-subagent workload averaged 18 min on single RTX Spark vs 8m48s on 3-device cluster
NVIDIA labels it unofficial, not a formal benchmark
Limitations: single scheduling policy, no VRAM/model warmness awareness
Supports RTX, DGX Spark, and Mac nodes
PAIR proxies existing Ollama and LM Studio endpoints with no agent harness changes required
NVIDIA demo: 5-subagent task took 18 min on single RTX Spark, 8 min 48 sec on 3-device cluster
NVIDIA labels demo unofficial, not a benchmark
Open source release
Known limitations: single scheduling policy, no VRAM awareness, no model warmness tracking
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: We look at NVIDIA Personal AI Router (PAIR), an open source virtual inference router that spreads local AI requests across the machines already on a home network. We cover how PAIR proxies existing Ollama and LM Studio endpoints so agent harnesses need no changes, and how its scheduler filters…