Control the AI Model on Your Mac from Your iPhone
LM Studio announced the LM Link feature, which allows you to use local artificial intelligence models on Mac via iPhone. Details are in our news.
LM Studio updated its well-known Mac application and Locally AI iOS application and introduced a new feature called LM Link. This innovation allows users to interact with large language models (LLM) running on Mac computers directly from their iPhone devices.
LM Link works by establishing an end-to-end encrypted connection between iPhone and Mac. In this way, users can benefit from the artificial intelligence capabilities on the Mac while maintaining the privacy advantages offered by local models.
Access to local models on Mac from iPhone with LM Link
LM Studio stands out as a tool that helps users discover the AI models best suited to their hardware and manage performance settings. The newly added LM Link feature brings this experience to portable devices, making it possible to use the local processing power on the Mac via the iPhone.
To use the feature, you need to sign in with an LM Studio account on both devices. The relationship runs on Tailscale mesh VPN infrastructure and the devices are not exposed to the public internet in any way.
LM Studio states that this system will not conflict with an existing Tailscale VPN installation and offers a completely independent structure. LM Link is compatible with all local models installed on Mac, including the Apple Intelligence base model.
LM Link, which is currently in the preview process, can be used free of charge. The company plans to introduce priced subscription options in addition to the free plan in the future.
User experience and technical details
In the first tests conducted with LM Link, controlling local models via iPhone offers a very productive experience. While it is observed that the connection can sometimes be quickly interrupted when the application is put in the background, the developers explain this situation as a situation arising from the structure of the relationship protocol.
The development team continues to work to reduce delay times and extend contact time. For users who want to try local models, this period is a very suitable time period, especially with the Mac support of models such as Gemma 4 with 12 billion parameters.
What do you think about the idea of managing local artificial intelligence models on Mac and using them with iPhone?
