NVIDIA ships open-source PAIR: turns idle home PCs into a local AI cluster, ~51% faster in demo
solyarisoftware · x · 2026-09-10
NVIDIA has released PAIR (Personal AI Router), free and Apache 2.0 open source: install it on computers you already own and it auto-discovers them, routing local AI work to whichever machine has capacity.
- Supported: GeForce RTX 20-series+, RTX PRO, DGX Spark/GB10, Apple M4+ Macs; Windows/Linux/macOS
- No exotic stack needed: works with Ollama, LM Studio, and existing OpenAI-compatible apps
- NVIDIA's demo (Hermes Desktop + Ollama, 5 subagents): one RTX Spark laptop took 18 minutes vs 8m48s across three machines — roughly 51% less completion time
Machines don't even need to be idle to participate.
More from Infra
- turbovec: Rust vector index fits 10M document vectors in 4GB RAM and outpaces FAISS — tom_doerr · 2026-09-10
- LM Studio 0.4.24 adds advanced llama.cpp argument overrides for GGUF model loading — solyarisoftware · 2026-09-10
- tszzl wraps up: efficiency gains only amplify hunger for hardware — tszzl · 2026-09-10
- Trimming MTP draft vocab to 47k boosts DGX Spark code decoding by 21.5% on same hardware — MaziyarPanahi · 2026-09-10
- Is 5 tokens/s usable for local LLMs? Redditor runs 27B model off an iGPU — Zombiecidialfreak · 2026-09-10
- Deep-Dive Speculative Decoding Blog Incoming: Drafter Training to vLLM Serving — auto_grad_ · 2026-09-10