l3ms: local AI that fits your Mac
Most AI apps have a ceiling: your Mac has to ask a server for permission to think.
l3ms removes that ceiling. It is a macOS app that manages local LLMs, and it is built around one rule: the model should fit the machine, and the machine should keep the data.
What it does
- Checks your hardware. l3ms reads your Mac's chip, memory, and GPU headroom.
- Recommends a model that fits. Not the biggest model on the shelf. The model that will actually run well on your machine, with an honest tradeoff between speed and quality.
- Downloads and runs it. One click. Inference runs locally through ollama, streaming responses with no uploads.
The result is a chat window that works like a regular app, with no account, no queue, and no data leaving your machine.
Why local
Three reasons, in order of importance:
- Privacy. Prompts and conversations stay on your Mac. No cloud, no account, no data collection. The cloud symbol in the icon is crossed out on purpose.
- Offline. Once a model is downloaded, l3ms works without an internet connection. On a train, in a cabin, or in the middle of a network outage.
- Ownership. The model is a file on your disk, the same way your music and photos are. You are not renting a service; you are using software.
Part of a larger suite
l3ms is the first app in a larger suite of local, offline-friendly apps (web, Mac, and mobile) developed by AIVI AI Services, an AI consulting firm based in Amsterdam. Every app in the suite follows the same rule: your data stays on your device, and the software runs without a subscription to a cloud service.
What is next
l3ms is free during early access. We want to see how people use it before deciding on licensing, so the app carries a donate button rather than a paywall for now.
If you have a Mac and a model you have been meaning to try, download l3ms, let it look at your hardware, and see what it recommends. The right model for your machine is probably smaller than you think.