Getting started

Install LM Mini, pick on-device, Wi‑Fi, or Mac Home, and send your first message in a few minutes.

Answer first

You are ready when a reply streams. Three ways in: a model on this device, LM Studio/Ollama on the same Wi‑Fi, or LM Mini Home on a Mac (Share with phone + optional chat sync).

Requirements

Phone: iOS 15.5+ or Android 8.0+. Mac Home is optional. You do not need a PC for on-device chat.

Pick a path

Path	Use when	Next
On this device	Travel, privacy, no PC	Download a model in the app
Same Wi‑Fi	A computer already runs LM Studio or Ollama	Point LM Mini at that computer
Phone + Mac	You bounce between devices	LM Mini Home → Share with phone → sync

You do not need a PC for on-device chat.

Path 1 — On this device (fastest first message)

Install LM Mini from the App Store, Google Play, or LM Mini Home on Mac.
On first launch, choose Beginner (simple defaults + a model that fits your RAM) or Power user (full Settings).
Pick Faster / Balanced / Best — the app matches your device. Tap to download.
Wait for the download (Lock Screen / notification shows progress on iPhone and Android).
Open a new chat and send “hello”.

If you skipped setup and land on an empty chat, Get Started offers the same download — or connect a computer instead.

Tip

Smaller quants (Q4 / 4-bit) fit phones. Bigger “Best” picks need more RAM. You can add more later under Settings → Models.

Path 2 — LM Studio or Ollama on Wi‑Fi

On the computer, start LM Studio (Developer → Start Server, Serve on Local Network, port 1234) or Ollama (often port 11434).
Load or pull a model on that computer.
In LM Mini: Settings → Server → add http://YOUR_LAN_IP:1234 (or :11434 for Ollama, :1337 for JAN AI).
Pick the loaded model in the chat header and send a message.

Both devices must be on the same Wi‑Fi. Guest networks and “AP isolation” block this.

Finding your IP

Mac — System Settings → Wi‑Fi → Details. Terminal: ipconfig getifaddr en0
Windows — Settings → Network → Wi‑Fi → Properties, or ipconfig (IPv4)
Linux — ip addr or hostname -I

If discovery is available in Settings, LM Mini can scan for LM Studio. Auth-protected servers still show up — paste the API token after you pick one.

Do not type localhost on the phone. That is the phone itself, not your PC.

Path 3 — Phone and Mac together

This is the everyday Home setup: chat on the phone, sit down at the Mac, keep going.

Install LM Mini Home on the Mac. Open it, then Share with phone.
On the phone: Settings → Remote Access → scan the QR.
When asked Keep chats in sync?, turn it on if you want threads and folders on both devices. Details: Phone and Mac sync.

Connect for Mac is deprecated through the end of 2026. New Mac setups should use Home. Windows and Linux still use Connect.

First-run extras

Find a model that fits — optional Arena race (about a minute). Skip if you already know.
Theme — you can change it later in Appearance.
Cloud API keys are optional. Add them later under cloud providers.

Providers & models — engines, catalog, cloud
Chat — folders, per-chat settings, context
macOS desktop — Home shell, local models, menu bar
Blog: LM Studio / Ollama — same story in plain language