mcmonkeyprojects/SwarmUI

SwarmUI – A Modular Web UI for AI‑Generated Images (and more)

What it is – SwarmUI (formerly StableSwarmUI) is an open‑source, web‑based interface that lets you run and control a variety of generative AI models from a browser. It focuses on image generation (Stable Diffusion, Flux, Krea 2, etc.) but also includes support for video and audio generation models. The UI is built to be fast, extensible, and friendly to both beginners and power users.

Key capabilities

• Multi‑model support – Plug‑in style back‑ends for Stable Diffusion, Flux, MiniMax H3, Wan, LTX‑2, ACE‑Step, and others. New models can be added via extensions.
• Generate tab – Simple point‑and‑click interface for quick image creation, with helpers like an image editor, auto‑workflow generation, and a Grid Generator for large batch renders.
• Comfy Workflow tab – Exposes the raw graph‑based workflow (similar to ComfyUI) for advanced users who want unrestricted control.
• GPU swarm – Designed to let many GPUs work together on the same request, enabling fast large‑grid generation.
• Extensible architecture – Users can write extensions (e.g., the upcoming native LLM‑assisted prompting) and install community‑made node packs.
• Cross‑platform – Runs on Windows, Linux, and Apple‑silicon Macs; Docker images are also provided.
• One‑click installers – Windows .bat and Linux .sh scripts automate dependency installation (Git, .NET 8 SDK, Python, required Python packages) and launch the server.
• Cloud‑ready – Ready‑made templates for Runpod and Vast.ai let you spin up a SwarmUI instance on a GPU cloud with minimal effort.

How to get started

1. Choose your platform – Windows (installer script), Linux (install‑linux.sh or manual), macOS (Apple‑silicon only), or Docker.
2. Run the installer – The script pulls the .NET runtime, creates a Python virtual environment, downloads model files, and starts a local web server (default http://localhost:7801).
3. Open the UI – Your browser will open the SwarmUI web UI. The first page walks you through model selection and basic configuration.
4. Generate – Use the Generate tab for quick prompts, or switch to the Comfy Workflow tab for custom node graphs.

Project status – The repository labels the current release as “Almost‑Release” (v0.9.8 Beta). Core features are stable, but a few items are still on the roadmap, most notably native LLM‑assisted prompting (currently only available via an extension).

Community & support

• Discord – Active chat for troubleshooting, feature requests, and announcements.
• GitHub Discussions – Feature‑announcement thread and issue tracker.
• Documentation – /docs folder contains a basic usage guide, advanced usage notes, Docker instructions, and motivation documents.

Legal & licensing

• SwarmUI itself is released under the MIT License (© 2024‑2026 Alex “mcmonkey” Goodwin).
• The UI can automatically install and interface with several third‑party projects (ComfyUI, AUTOMATIC1111’s web UI, YOLOv8, etc.) that carry GPL, AGPL, Apache‑2, or other licenses. Users should be aware that those components’ licenses may affect redistribution.
• Model files are not included; they retain their own licenses.

Why it matters – While many web UIs for Stable Diffusion exist, SwarmUI aims to combine the ease‑of‑use of a single‑click generator with the flexibility of a full workflow editor, and it adds the ability to harness multiple GPUs as a “swarm” for high‑throughput batch jobs. This makes it a practical tool for artists, researchers, and hobbyists who want a unified front‑end for a growing ecosystem of generative AI models.