LM Studio launched LM Link today, a new feature that connects its Mac app with the Locally AI iOS app to let users interact with local language models from their iPhones. The update allows seamless communication between devices without exposing data to the public internet, maintaining the privacy benefits of running LLMs locally on Mac hardware.

LM Link requires users to create an LM Studio account and sign in on both devices. Once activated, it establishes an end-to-end encrypted connection using custom Tailscale mesh VPNs. This approach keeps all data private and secure while avoiding interference with existing Tailscale VPN setups. The feature works with any model installed on the Mac, including Apple's built-in foundation model.

The author tested LM Link for several days and found it useful for accessing local LLMs on the go. However, the secure connection occasionally drops when the iPhone app runs in the background for extended periods. Developers acknowledge this limitation and are working to improve reconnection speed and connection stability.

During the Preview period, LM Link remains free. Paid plans will launch afterward, though specific pricing details haven't been revealed. For users already running local models on their Macs, this integration provides a convenient way to extend access to mobile devices while preserving data privacy.