I’ve played around a lot with Locally AI, an app for running local models on your phone. Every time a phone-sized model comes out I open Locally, try out the model, and forget about the app until I’m on a flight without WiFi. Recently, though, Locally was acquired by Element Labs, the makers of LM Studio. Since then, Locally has added support for LM Link — a feature that allows you to access your local models remotely.
LM Studio is an app for running local models on your computer. Since your computer has more RAM and processing power than your phone, you can run more powerful models if you connect back to your computer. How capable a model you can run depends on your computer’s specs, but many common setups will let you run genuinely useful models like Qwen 3.6, Gemma 4, or Muse Glimmer.
What stands out about LM Link is how seamlessly it connects two devices running LM Studio. If I have LM Studio open on my MacBook Air, I can use the dramatically better GPU and more RAM my MacBook Pro has to run more powerful models there. I played around a bit with the feature when it first came out, but it became much more compelling once Locally AI added LM Link support. Now it’s incredibly easy to connect to my MacBook Pro sitting at home, so I can use a dramatically better local model whenever I’m out.
OpenAI doesn’t offer the privacy I want for my most sensitive questions, so when I don’t need a data center’s worth of intelligence, I’m happy to use a local model. Most of the time when I’m out of the house I still have an internet connection, so this solution provides a sensible alternative. When I need more intelligence than my phone can offer but want to feel certain nobody will read my Riker Googling-style questions, I call out to my MacBook Pro running LM Studio. Now I have the best of both worlds, with only a hint of compromise.