The app store Stable Diffusion deserved, not the one it got
A C# desktop app that installs, updates, and corrals two dozen competing AI image generators so you don't have to.

What it does Stability Matrix is a cross-platform desktop package manager—built in Avalonia UI—that wrangles the sprawling ecosystem of Stable Diffusion front-ends. One click installs or updates Automatic1111, ComfyUI, Fooocus, InvokeAI, Forge, SD.Next, Kohya’s GUI, FluxGym, and roughly a dozen others. It bundles its own Git and Python runtimes, so your system stays clean. Everything lives in a portable data directory you can drag to another drive intact.
The interesting bit
The “Inference” tab is the twist: a built-in generation UI with formal-language auto-completion, dockable panels, and .smproj project files. Generated images embed ComfyUI node graphs and A1111-compatible metadata, so you’re not locked into Matrix’s own format. It’s an attempt to be both the installer and the daily driver.
Key highlights
- One-click installs for 20+ packages including reForge, StableSwarmUI, VoltaML, SDFX, OneTrainer, and CogVideo via CogStudio
- Shared model directory with drag-and-drop importing; auto-fetches CivitAI metadata and preview thumbnails
- Integrated model browser for CivitAI and HuggingFace with pause/resume downloads
- Plugin/extension manager for A1111, ComfyUI, SD Web UI-UX, and SD.Next
- Syntax-highlighted terminal emulator with launch argument editor and environment variable config
- Localized in 12 languages via community contributors
Caveats
- macOS support is Apple Silicon only; no Intel Mac builds mentioned
- Binaries ship under a separate End User License Agreement, while source is AGPL-3.0—a license split worth noting
- The sheer breadth of supported packages suggests maintenance load; how quickly new releases get packaged is unclear from the README
Verdict Grab this if you’re tired of maintaining separate Python environments for every new SD fork, or if you hop between ComfyUI and Fooocus and want one model folder. Skip it if you’re already happily married to a single frontend and don’t need the abstraction layer.