Lemonade

Lemonade helps users discover and run local AI apps by serving optimized LLMs, images, and speech right from their own GPUs and NPUs.

Installation

Lemonade is available in the AUR:

• Install lemonade-server^AUR for the LLM server and router.
• Install lemonade-desktop^AUR for an electron GUI application.

Recommended dependencies

• Install xrt and xrt-plugin-amdxdna for AMD NPU support.
• Install fastflowlm for running LLMs on the NPU.

Note For kernels older than 7.0, you also need to install the amdxdna-dkms^AUR dkms driver in order for the NPU to be usable.

Usage

Systemd service

When Lemonade Server is installed, it includes a systemd service called lemonade-server that allows you to manage the server process using standard systemd commands.

Common Commands

# Start the service 
sudo systemctl start lemonade-server

# Stop the service
sudo systemctl stop lemonade-server

# Restart the service
sudo systemctl restart lemonade-server

# Check service status
sudo systemctl status lemonade-server

# Enable automatic startup on boot
sudo systemctl enable lemonade-server

# Disable automatic startup on boot
sudo systemctl disable lemonade-server

# View recent logs (follow mode)
sudo journalctl -u lemonade-server -f

Note Service Features:

• Automatic restart: The service automatically restarts if the server crashes
• User isolation: Runs under the unprivileged lemonade user for security
• Security hardening: Includes systemd security options like ProtectSystem=full, ProtectHome=yes, and NoNewPrivileges=yes

CLI usage

To see the available models and download them:

$ lemonade-server list
$ lemonade-server pull Gemma-3-4b-it-GGUF

To see the backends available on your PC:

$ lemonade-server recipes

To run and chat with Gemma 3:

$ lemonade-server run Gemma-3-4b-it-GGUF

Configuration

The Lemonade Server systemd service reads configuration in this order:

• /etc/lemonade/lemonade.conf
• /etc/lemonade/conf.d/*.conf

Use /etc/lemonade/lemonade.conf for base settings, add local overrides as numbered drop-ins under /etc/lemonade/conf.d/, and keep secrets like LEMONADE_API_KEY in /etc/lemonade/conf.d/zz-secrets.conf.

Edit the base configuration file:

$ sudo nano /etc/lemonade/lemonade.conf

Create a drop-in override file:

$ sudo nano /etc/lemonade/conf.d/50-local.conf

Secrets, like LEMONADE_API_KEY, are defined in /etc/lemonade/conf.d/zz-secrets.conf:

$ sudo nano /etc/lemonade/conf.d/zz-secrets.conf

Using /etc/lemonade/conf.d/ for local overrides keeps them separate from the package-provided base config and is the recommended path for persistent customization.

After making changes to the configuration files, restart the service for changes to take effect:

$ sudo systemctl restart lemonade-server

Graphical User Interface

The graphical user interface is both available in lemonade-server^AUR as a web-app, and in lemonade-desktop^AUR as an electron desktop application.

This app provides a native desktop experience for managing models and chatting with LLMs running on lemonade-router. It connects to the server via HTTP API and offers a modern, resizable panel-based interface.

Note Key Features:

• Model management (list, pull, load/unload)
• Chat interface with markdown/code rendering and LaTeX support
• Real-time server log viewer
• Persistent layout and inference settings
• Custom frameless window with zoom controls