The Birth Story of LM Mini (Or: How a Pet Project Escaped the Lab)
Hello, and welcome to the birth story of LM Mini.
What started as a pet project slowly turned into something I couldn't keep to myself. Back then it was just me and my phone, chasing a simple idea: local AI on mobile should feel easy, and it should stay private. The early versions were... a lot. Complicated. Built for me, by me, with buttons in places only I knew about.
Most people don't live and breathe AI, and honestly, they shouldn't have to. So I kept simplifying. Fewer steps. Clearer screens. Less "what does this setting even mean?" That's the whole point: LM Mini used to be the kind of thing only a tinkerer would love. I turned it into something you can actually hand to someone and say, just open it and start chatting.
At some point I stopped building it only for me and started building it for anyone who'd ever thought: I want AI on my phone, but I don't want my phone to become someone else's product.
So, what is LM Mini?
In the least jargony way I can manage: LM Mini is a chat app that puts AI on your phone, without sending your life story to a cloud server.
In practice, that usually looks like one of three things. Some people download a model and chat offline, anywhere and on the go. Some already run LM Studio or Ollama on a PC and just want a good mobile chat app on the same Wi‑Fi. And some use LM Mini Connect when they're away from home but still want the big model on their desk, not a tiny one in their pocket. Same app, different day. I use all three depending on where I am, which I guess is the whole point.
You can run models directly on your iPhone, iPad, or Android. Or connect to LM Studio, Ollama, or oMLX on your computer at home. Or chat from anywhere using LM Mini Connect, the desktop companion I built. LM Mini Connect is the desktop app that links your phone to LM Studio or Ollama at home. When you pair devices, you scan a QR code and the connection is encrypted with a key only you control, so your chats aren't sitting open on the internet and we can't read them either. No VPN setup. No opening ports on your router. Just your devices talking to each other, privately.
The encryption part matters more than it sounds. When you pair with a QR code, you're not trusting us to "keep your data safe" in the abstract. You're using a key that lives with you, so the connection between your phone and your computer is meant to be readable only on your side. Not by us, not by someone on the coffee shop Wi‑Fi, and not as plain text bouncing around the internet. Remote access is a Pro feature because that relay and upkeep aren't free to run, but the idea is the same: your chat stays yours. If you've ever wanted LM Studio on your phone without duct taping five apps together, this is the problem I was trying to solve.
Ultimately, I built it because I wanted local AI to feel normal and private. Not a weekend project for people who enjoy pain. Not something you need a PhD and three Reddit tabs open to use. Just... chat. On your phone. Yours. The kind of thing you can show someone without apologizing for the UI or explaining what a GGUF is before they've had coffee.
That's it, really. The fancy stuff is there if you want it: personas, group chats, voice mode, themes, tools, image generation, the whole toy box. But the core idea is simple: private AI that lives on hardware you actually own.
LM Mini isn't made by the LM Studio or Ollama teams. I built it independently, to work with the setup you already have. Think of it as the mobile half of a very nerdy friendship that actually answers back.
If you've been curious about local AI but didn't know where to start: download the app, pick a model, say hello. Worst case, you learn something. Best case, you stop copy-pasting into a browser tab like it's 2023.
More posts coming every few days: setup guides, feature spotlights, and the occasional rant about RAM. For now: welcome. Glad you're here.