This release introduces cross-distribution installation support, multimodal model (vision) integration via --mmproj, a real-time launch command preview, and expanded documentation for Ubuntu/Debian and Fedora users.
What’s New:
- Universal Installer (
install.sh): A new installation script that automatically detects the Linux distribution (Ubuntu/Debian, Fedora, Arch Linux), installs required system dependencies, creates a Python virtual environment, installs Python dependencies, generates a~/.local/bin/llamatraylauncher script, and creates a~/.local/share/applications/llamatray.desktopentry for application menu integration. - Desktop Integration (
.desktopfile): LlamaTray now appears in the application menu on GNOME, KDE, XFCE, and other desktop environments after runninginstall.sh. - Multimodal Model Support (
--mmproj): A dedicated “mmproj File” field has been added to the Advanced Settings panel, featuring a file picker dialog (.gguffilter). The selected mmproj path is automatically appended as--mmproj /path/to/fileto the launch command, enabling vision-capable models such as Qwen3-VL and Gemma4 multimodal. - Launch Command Preview: A new read-only “Launch Command Preview” panel has been added below the Advanced Settings. It displays the full
llama-servercommand in real time as parameters are changed, providing complete transparency before starting the server. - Multi-Distribution README: Installation instructions for Ubuntu/Debian and Fedora have been added to
README.mdalongside the existing Arch Linux section.