LM Mini Documentation

LM Mini is a mobile chat client for LM Studio. It lets you chat with locally running large language models from your iOS or Android device.

🚧 Documentation in Progress

We're actively building out the documentation. Check back soon for detailed guides on every feature.

Quick Start

Install LM Studio
Download and install LM Studio on your computer. Load a language model.
Start the API Server
In LM Studio, go to the Developer tab and start the local server. The default address is localhost:1234.
Install LM Mini
Download LM Mini from the App Store or Google Play.
Connect
Open LM Mini and enter your computer's local IP address (e.g., 192.168.1.100:1234). Make sure both devices are on the same Wi-Fi network.
Start Chatting
Select a model and start a conversation. That's it!

Requirements

Component	Requirement
LM Studio	v0.3.0 or later
iOS	iOS 15.5 or later
Android	Android 8.0 (API 26) or later
Network	Same Wi-Fi network (or use LM Connect for remote access)

Chat Interface

LM Mini provides a full-featured chat interface with:

Markdown rendering with syntax highlighting
LaTeX math equation support
Conversation folders and organization
Custom themes, avatars, and backgrounds
Adjustable font size
Export conversations to JSON or text

Detailed documentation coming soon.

Model Management

Browse, download, and manage models directly from LM Mini:

View all loaded and available models
Download new models from LM Studio's catalog
Switch models mid-conversation
Configure model load parameters (context length, flash attention, etc.)

Detailed documentation coming soon.

Voice Chat

LM Mini supports hands-free voice conversations:

Speech-to-text input using native recognition
Text-to-speech output with adjustable speed and pitch
Continuous conversation mode
Multiple TTS providers

Detailed documentation coming soon.

Tool Calling & MCP

Extend your AI with real-world capabilities:

Web search integration
MCP (Model Context Protocol) server support
Ephemeral MCP servers via HTTP
Integrated MCPs from LM Studio's mcp.json

Detailed documentation coming soon.

Image Generation

Connect to AUTOMATIC1111 / Stable Diffusion WebUI to generate images:

Generate images from chat prompts
Full control over parameters (steps, CFG scale, sampler, etc.)
High-res upscaling support

Detailed documentation coming soon.

Remote Access

Access your LM Studio from anywhere using LM Mini Connect:

No port forwarding or VPN needed
Secure encrypted relay
Simple QR code pairing
Works across networks

See the LM Connect Setup Guide for details.

Cloud Providers

LM Mini Pro users can connect to cloud AI providers:

OpenAI (GPT-4o, o1, etc.)
Anthropic (Claude)
Google (Gemini)
Groq, Mistral, OpenRouter, and more

Detailed documentation coming soon.

LM Connect Setup Guide

Download LM Mini Connect
Download the desktop companion app for your platform (macOS, Windows, or Linux).
Enter LM Studio Address
Set the LM Studio address (usually localhost:1234). If authentication is enabled, enter the API token.
Click Connect
The app will connect to the relay server and show a QR code.
Scan in LM Mini
Open LM Mini on your phone → Settings → Remote Access → Scan QR Code.
Done!
Your phone is now connected to LM Studio via the secure relay. This works from any network.

Troubleshooting LM Connect

Connection fails immediately

Make sure LM Studio is running and the API server is started. The default address is localhost:1234. If you changed the port, update it in LM Mini Connect.

QR code scan doesn't work

Ensure you have LM Mini Pro to use remote access. The QR code scanning feature requires a Pro subscription.

High latency

The relay server adds a small amount of latency. For the best experience, ensure both your computer and phone have a stable internet connection.

Model Parameters

Fine-tune model behavior with these parameters:

Parameter	Range	Description
Temperature	0.0 — 2.0	Controls randomness. Lower = more focused, higher = more creative.
Top P	0.0 — 1.0	Nucleus sampling. Limits to tokens with cumulative probability ≤ top_p.
Top K	1 — 100	Limits to the K most likely tokens at each step.
Min P	0.0 — 1.0	Minimum probability threshold relative to the most likely token.
Max Tokens	1 — 32768	Maximum number of tokens in the response.
Repeat Penalty	1.0 — 2.0	Penalizes repeated tokens to reduce loops.

System Prompts

Customize the AI's behavior with system prompts. You can create and save a library of prompts for different use cases.

Detailed documentation coming soon.

Appearance

Customize LM Mini's look and feel:

Light, dark, or system theme
Custom chat background images
User and assistant avatars
Adjustable chat font size (10pt — 24pt)
Per-conversation bubble colors

Detailed documentation coming soon.