A simple GUI wrapper for llama.cpp built with JavaFX.

This project is a simple GUI wrapper for llama.cpp. Currently it supports the llama-server tool only.

After downloading or building the application, simply run the binary file llama-fx, no installation needed.

Set the path to your llama.cpp folder (you should see llama-server in this folder). Then you can set parameters for the llama server and pick a GGUF model.

Once your configuration is ready, click Launch to start a llama-server instance.

You can click Open Web UI to open a new tab in your browser to the llama server's web UI.

You can also save configurations and load them, which is helpful for switching between different models and settings, or for launching multiple instances.

This project was built with the following:

• Java 21
• IntelliJ
• Maven
• JavaFX

Clone the repository, and then open the project in IntelliJ as a Maven project.

You can run the application either with mvn javafx:run or by running the main class MainApplication.java normally.

Builds are created using jlink and jpackage Maven plugins. You can see the result in /target/dist where you'll find the binary file llama-fx, and /app and /runtime folders next to it.

Run one of the Maven commands below depending on your platform:

mvn clean javafx:jlink jpackage:jpackage@win

mvn clean javafx:jlink jpackage:jpackage@linux

mvn clean javafx:jlink jpackage:jpackage@mac

• llama-fx does not come with llama.cpp included. You have to download llama.cpp yourself and point llama-fx to it.
• llama-fx does not come with any models either. You pick a GGUF model you already have for llama-fx to pass to llama.cpp when launching.