What is Mignon UI?
Mignon UI is a private, offline-first client for local AI roleplay and storytelling. It allows you to create multi-character lobbies, customize aesthetics with retro themes, and maintain long-term memory across sessions.
Local & Offline-First
Roleplay with AI on your own terms.
Chat with multiple AI characters in styled lobbies. Zero telemetry, custom themes, and complete control over your models.
Designed for Immersion
A clean, distraction-free desktop client engineered to get out of the way of your stories.
Dynamic Multi-Bot Lobbies
Chat with multiple characters in the same room. Characters take turns naturally, talking to you and each other based on their personality and context. You can also play as any character card.
Private & Offline-First
Your logs, characters, and API keys are stored in a secure local database directly on your device. Zero telemetry.
Tavern Card Import
Drag and drop standard V2 PNG character cards or JSON cards to import characters instantly.
Smart Story Memory
Long roleplays stay consistent. The client summarizes key story moments into chapters and automatically retrieves relevant context so characters never lose track of the plot.
Download Mignon UI
Get the latest packages for your platform. Fully offline, standalone client.
Windows
Supported on Windows 10 and 11. Run completely locally or connect to local/cloud backends.
⚠️ Security Note: Since the app is newly compiled and unsigned, Windows SmartScreen will show a warning. Click "More info" and then "Run anyway" to proceed.
macOS
Universal binary supporting both Apple Silicon (M1/M2/M3) and Intel Core Macs.
⚠️ Security Note: Right-click (or hold Control and click) the app icon in Applications, select Open, and click Open again in the Gatekeeper warning box.
Linux
Compatible with Ubuntu, Debian, Arch Linux, and other major distributions.
🔧 AppImage Tip: Make sure to set executable permissions before double-clicking:
chmod +x Mignon-UI*.AppImage.
Connect Your Local Engine
Connect to any local or cloud provider. Running locally guarantees unlimited, free, and private chat.
Ollama Integration
Run open-weights models locally in one command. Mignon UI links automatically.
Terminal
ollama run llama3.1
Default Endpoint: http://127.0.0.1:11434/v1
Kobold.cpp Optimization
Optimized for limited hardware (e.g. 6GB VRAM GPUs). Enable SmartCache and ContextShift.
Tuning
Enable: ContextShift + SmartCache + Q4 KV Cache
Default Endpoint: http://127.0.0.1:5001/v1
Frequently Asked Questions
Everything you need to know about Mignon UI and local AI roleplay.
Is Mignon UI free and private?
Yes, Mignon UI is 100% free, open-source (AGPL-3.0), and offline-first. Your chats, character cards, and API keys are stored locally on your device with zero telemetry.
Which local AI engines are supported?
Mignon UI works out of the box with Ollama, Kobold.cpp, LM Studio, LocalAI, vLLM, and any server exposing an OpenAI-compatible API endpoint.
Can I import Tavern character cards?
Yes! You can drag and drop standard V2 PNG character cards or JSON cards to import characters, personalities, scenarios, and greeting messages automatically.