NVIDIA PAIR: free Apache-2.0 software for routing local AI inference

The offer

Download NVIDIA PAIR for Windows, Linux or macOS and route independent AI requests across compatible computers on your local network. The software is free under Apache-2.0, including for users in India; you supply the hardware and models, and separate engine, application or provider terms and charges may apply. PAIR does not pool GPU memory or split a model across computers, and no hosted AI subscription is included.

Who can claim it

• Worldwide Apache-2.0 licence; no India-specific exclusion found.
• Use a supported Windows, Linux or macOS system.
• Serving nodes need a compatible inference engine and enough memory for the selected model.

How to claim it

1. Open the official page and select Download PAIR.
2. Download and install the package matching your operating system and architecture.
3. Install or enable Ollama, LM Studio or llama.cpp and download a compatible model.
4. For multiple computers, install PAIR on each and pair them on the same trusted local network.
5. Copy the local URL from Endpoints and configure your compatible AI application to use it.

Good to know

No expiry is stated for the downloaded Apache-2.0 software; this is not a subscription trial.

Verified release: v1.0.0. Hardware, electricity and model storage are not included. Engines, models and connected applications have separate terms and may have separate charges. PAIR routes each request to one node; it does not pool GPU memory or split a model across machines. Windows ARM support is experimental.