Documentation

LM Mini docs

Practical guides for on-device AI, LM Studio & Ollama, phone ↔ Mac sync, Personas, group chat, voice, tools, and LM Mini Home — written so you can find the setting and finish the task.

Quick start Read the blog

Answer first

LM Mini is a private AI chat app for iPhone, Android, and Mac. Run models on-device, connect over Wi‑Fi to LM Studio/Ollama, or use LM Mini Home → Share with phone for encrypted remote access and chat sync. Connect for Mac is deprecated through the end of 2026. Use the topic cards below — each page is a focused guide, not one endless scroll.

Start here

Getting started

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

Setup

Providers & models

Switch between on-device GGUF/MLX, LM Studio, Ollama, JAN AI, oMLX, Unsloth, and cloud — plus the in-app catalog, what's free vs Pro, downloads, and reasoning.

Remote

Remote access

Reach home models from anywhere. On Mac, use LM Mini Home → Share with phone. Connect for Mac is deprecated through the end of 2026. Windows and Linux use LM Mini Connect.

Remote

Phone and Mac sync

Keep chats, folders, and the personas you pick in sync between your phone and LM Mini Home so you can continue on either device.

Core

Chat interface

Streaming chat, folders, per-chat settings, context fit, attachments, and the composer in LM Mini.

Core

Personas

Saved AI characters with avatar, prompt, model, Kokoro voice, and optional image workflows — plus memory and custom params with Pro. Includes built-in Mini.

Core

Group chat

Multi-model AI rooms with Personas, turn modes, parallel streaming, scenarios, and group voice in LM Mini (Pro).

Core

Voice

TTS, speech-to-text, hands-free continuous conversation, Whisper, Mac system speech, and Kokoro on-device or on LM Mini Home / Connect.

Power

Tools, MCP & Pro Search

Tool calling, Pro Search, SearXNG, ephemeral MCP servers, LM Studio integrated MCPs, and code sandbox in LM Mini.

Power

Image generation

Generate images with AUTOMATIC1111, ComfyUI workflows and LoRAs, or on-device Core ML Stable Diffusion in LM Mini.

Power

Arena

Race up to four models on the same prompt to compare speed and answer quality in LM Mini Arena.

Desktop

macOS desktop

LM Mini Home is the Mac app — nav rail, local models, Share with phone, and chat sync with your phone. Connect for Mac is deprecated through the end of 2026.

Reference

Settings & toggles

Map of LM Mini settings — providers, Home sync, model parameters, appearance, voice, widgets, Shortcuts, and the Pro settings (param presets, App Lock, memory, backup).

Help

Troubleshooting

Fix LM Mini connection errors, Home sync, context-full chats, voice, tools, and common setup issues.

Open source

Build from source

Clone, build and run the open-source LM Mini core on Android, iPhone, iPad and Mac.

Common questions

How do I connect LM Mini to LM Studio?

Install LM Studio, start the Developer API server with Serve on Local Network, open LM Mini, enter your computer's LAN IP and port (e.g. http://192.168.1.100:1234), then pick a loaded model.

What is on-device AI in LM Mini?

On-device AI runs GGUF or MLX models directly on your phone or Mac so chat works offline without a PC server. Free includes four small starter GGUF models (1–1.7B) and one imported file; LM Mini Pro unlocks the full catalog, including bigger and MLX models. The welcome flow and Get Started sheet suggest Faster / Balanced / Best picks for your RAM.

How does remote access work?

Remote access is an LM Mini Pro feature. On Mac, LM Mini Home Share with phone pairs via QR with an encryption key only you control. Connect for Mac is deprecated through the end of 2026. Windows and Linux use LM Mini Connect.

How do I keep chats in sync between phone and Mac?

Pair the phone with LM Mini Home, then enable Sync chats with Home under Settings → Remote Access (or accept Keep chats in sync? after pairing). Conversations and folders stay aligned, and Sync personas copies the personas you pick, so you can continue on either device. Pairing needs LM Mini Pro.

What replaced Connect for Mac?

LM Mini Home. Connect for Mac keeps working through 31 December 2026. Windows and Linux Connect remain current.

What are Personas?

Personas are saved AI characters with name, avatar, system prompt, preferred model/provider, Kokoro voice, and optional ComfyUI workflow. With Pro they can also keep memories and custom params. New installs include a Default persona and Mini.

Does LM Mini support ComfyUI LoRAs?

Yes. In Settings → Image Generation choose ComfyUI, set your server URL, pick or paste a workflow, and use %LORA% / %LORA_WEIGHT% placeholders when needed.

How do I lock LM Mini with a PIN?

Pro users can open Settings → Data → App Lock, set a 4- or 6-digit PIN, and choose how long the app can stay closed before it asks again. The PIN stays on this device and is never synced. Face ID, Touch ID, or fingerprint can unlock it. If you signed in with Apple, Google, or email while the app was unlocked, you can recover a lost PIN from the lock screen: sign in again and confirm the verification email, then choose a new PIN. Without a linked account, reinstalling is the only way to clear the lock.

Is LM Mini open source?

The core app is. It is MIT-licensed on GitHub at github.com/slim-elephant/lm-mini. Pro features ship only in the official App Store, Google Play, and Mac builds. See lmmini.com/open-source.html and the Build from source guide.

What is JAN AI in LM Mini?

JAN AI is a free local provider. Start the JAN app, add it in Settings → Server (default http://LAN-IP:1337), and pick a model JAN has loaded. Do not use localhost on a phone.

Does LM Mini work with Unsloth Desktop?

Yes, for free since 1.9.10. Start Unsloth Desktop on your computer, add it in Settings → Server as http://LAN-IP:8888 with your sk-unsloth-… API key, and pick a model. Model Parameters → Context Length reloads the model at that size.