Open-source project
mcmonkeyprojects/SwarmUI avatar
mcmonkeyprojects/SwarmUI

SwarmUI: a modular Stable Diffusion web UI that puts ComfyUI under a Generate tab

SwarmUI (formerly StableSwarmUI), A Modular Stable Diffusion Web-User-Interface, with an emphasis on making powertools easily accessible, high performance, and extensibility.

4,613 stars461 forksC#MIT

At a glance

What is it?
SwarmUI is a C# web front end for AI image and video models that keeps a beginner-friendly Generate tab on top of a raw Comfy workflow graph. It is free, MIT licensed, and still labelled Almost-Release by its own README.
Who is it for?
Adopt SwarmUI if you want one interface that serves both a beginner clicking through a Generate tab and an advanced user who needs the raw Comfy graph, and you are willing to accept an Almost-Release project whose installer sometimes needs a second run. Do not adopt it if your hardware or workflow depends on something the README does not cover, or if you need a stable release rather than a Beta.
Can I use it commercially?
Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
Is it still maintained?
Yes. The repository last received commits 6 days ago.
What is it written in?
Mainly C#, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on September 28, 2026, and from our analysis. They are not legal advice.

Editorial analysis

What SwarmUI is for, and who it is actually aimed at

The README describes SwarmUI as a modular AI image generation web user interface, formerly named StableSwarmUI, with an emphasis on making power tools easily accessible, high performance, and extensibility. It supports AI image models such as Krea 2, Stable Diffusion and Flux, AI video models such as MiniMax H3, Wan and LTX-2, and some AI audio models such as ACE-Step.

The stated audience is two groups at once. Beginners are pointed at the primary Generate tab, which the README says makes it easy to generate anything with a variety of features. Advanced users are pointed at the Comfy Workflow tab, described as the unrestricted raw graph, with the README arguing they will still return to the Generate tab for the image editor, auto-workflow-generation and the Grid Generator power tool.

That dual audience is the whole design thesis. Most web UIs for diffusion models pick a side: either a simplified form that hides the sampler graph, or a node editor that assumes you know what a latent is. SwarmUI's bet is that the same server can expose both, because the underlying execution engine is Comfy and the Generate tab is a layer that writes workflows for you. Whether that bet pays off depends on how often you need to drop below the layer, which is exactly where the project's own documentation is thinnest.

How the Generate tab and the Comfy graph coexist

The repository layout tells you the shape of the system. There is a src/ directory holding the C# server, a Desktop/ directory for a desktop app mode, a colab/ directory with a notebook, a launchtools/ directory with install and launch helpers, and launch scripts for Windows, Linux and macOS at the top level. The topics list puts csharp, javascript and python side by side, which matches a C# backend serving a JavaScript front end while driving Python model code.

The README states that the Generate tab performs auto-workflow-generation. That is the mechanism that matters: you fill in a form, and the server emits a Comfy workflow from it. The Comfy Workflow tab then shows you the raw graph. Nothing in the README describes a two-way sync between the two, so the practical question for a user is whether a graph you hand-edit survives the next time the Generate tab regenerates a workflow. The documentation does not answer that, and it is the first thing I would test.

Releases are numbered as Betas, most recently 0.9.8-Beta on 2026-02-06, preceded by 0.9.7-Beta on 2025-08-25 and 0.9.6-Beta on 2025-04-15. The README labels the project Almost-Release and lists one key feature target not yet implemented: LLM-assisted prompting, which it says exists as an extension but should be natively supported.

Installing SwarmUI on Windows and taking the first generation

The Windows path is the one the README documents in most detail. It warns that Windows 10 users may need to install git and the .NET 8 SDK manually first, while Windows 11 automates that. You download the install-windows.bat file from the 0.9.8-Beta release, store it somewhere you want to install at, and the README explicitly says not to use Program Files. Then you run it.

Two caveats are stated plainly in the README. The installer sometimes closes without going further, and running it again fixes that, marked with a TODO to fix it. It also places a desktop icon you can use to relaunch the server later. When the installer completes it launches the server and then shows the install UI.

If you prefer to do it by hand, the README gives an alternate manual path. It says the current version targets .NET 8 but a future version will target .NET 10, so install both SDKs, then clone and launch:

bash
git clone https://github.com/mcmonkeyprojects/SwarmUI
cd SwarmUI
./launch-windows.bat

On Linux the prerequisites are git and python3, with pip and venv where the distribution does not bundle them. The README is specific about versions: you want Python 3.11 or 3.12, 3.10 should also work, 3.13 might work, and you should not use 3.14 or later. It suggests checking that the interpreter resolves before going further:

bash
python3.11 -m pip --version

The Linux easy install downloads a script and runs it. Note that the README itself flags the download link as pointing at the 0.6.5-Beta release even though the current release is 0.9.8-Beta, with the parenthetical that the link is still current:

bash
wget https://github.com/mcmonkeyprojects/SwarmUI/releases/download/0.6.5-Beta/install-linux.sh -O install-linux.sh
chmod +x install-linux.sh
./install-linux.sh

The script installs everything and eventually opens the webpage in your browser. From there you follow the on-page install instructions, and the README says to be patient because some install processing takes a few minutes while models download. The Basic Usage Guide in the docs folder is where it sends you next.

The installer's rough edges and the cases where SwarmUI is the wrong pick

The most concrete limitation is written into the install instructions themselves: the Windows installer sometimes needs to be run twice. That is a small thing, but it is a signal about the project's polish level, and it sits next to a TODO asking for a self-contained .msi or .exe pre-installer that does not exist yet.

The second limitation is version pinning. The README states that Python 3.14 or later should not be used, and that the project currently targets .NET 8 while a future version will target .NET 10. If your environment is pinned to a newer Python, or you are building on a machine where you cannot install two .NET SDKs, you are outside the documented path.

The third is the LLM prompting gap. The README lists LLM-assisted prompting as a key feature target not yet implemented, with an extension available in the meantime. If native LLM control is the reason you are looking at SwarmUI, you are adopting an extension, not a feature.

Where it is the wrong tool: if you only ever want the raw graph and nothing above it, the Generate tab and its auto-workflow-generation add a layer you will not use, and a plain ComfyUI install is a smaller surface. If you need a frozen, stable release rather than a Beta cadence, the release names tell you what you are getting. And if your hardware or model family is not in the README's list, the documentation is silent, not reassuring.

SwarmUI compared with ComfyUI: the difference is the layer above the graph

ComfyUI is the obvious comparison, and the repository makes it unavoidable: ComfyUI is a listed topic, and the README describes SwarmUI's advanced tab as giving you the unrestricted raw graph, which is what a ComfyUI user already has by default.

The difference in approach is where the form lives. In ComfyUI, the node graph is the interface, and convenience comes from custom nodes and saved workflows. In SwarmUI, the node graph is one tab among several, and the default interface is a Generate form that emits a workflow for you. That means a new user can produce an image without learning node wiring, and an experienced user can still open the graph when the form does not cover the case.

The cost of that design is the layer itself. Auto-workflow-generation has to guess, and guesses fail in ways that are harder to debug than a graph you built yourself, because the graph is generated rather than authored. The README's own framing acknowledges this tension by saying advanced users will still come back to the Generate tab for convenience features, which is an admission that the two tabs are not equivalent and that the form does something the graph does not.

A second practical difference is packaging. SwarmUI ships install scripts for Windows, Linux and macOS, a Colab notebook, and third-party templates for Runpod and Vast.ai. ComfyUI's distribution story is different, and if you have already built your workflow around it, moving means re-establishing your model paths and your extensions rather than importing them.

Maintenance cadence, licence, and what an upgrade costs you

The repository is not archived, and the last push was on 2026-09-09, which is recent enough that the project is being touched between releases. Releases, however, are slower and explicitly labelled Beta: 0.9.6-Beta on 2025-04-15, 0.9.7-Beta on 2025-08-25, and 0.9.8-Beta on 2026-02-06. If you track releases rather than commits, expect gaps of several months.

The upgrade mechanics are visible in the repository structure. There are separate launch scripts for development and normal use, including launch-windows-dev.ps1 and launch-linux-dev.sh alongside launch-windows.bat and launch-linux.sh. That split suggests the normal launch path is what you run day to day, and the dev path is for working on the code. The README does not document a rollback procedure if an upgrade breaks your setup, so a working install you can restore is worth having before you update.

The licence is MIT, stated in the repository and in the LICENSE.txt file at the top level. That is permissive: it allows commercial use and modification, and it requires that the copyright notice and permission notice be included in copies or substantial portions. It does not grant patent rights and it comes with no warranty. That is a description of the licence text, not legal advice; if your use is unusual, read LICENSE.txt and talk to a lawyer.

Editorial conclusion

Adopt SwarmUI if you want one interface that serves both a beginner clicking through a Generate tab and an advanced user who needs the raw Comfy graph, and you are willing to accept an Almost-Release project whose installer sometimes needs a second run. Do not adopt it if your hardware or workflow depends on something the README does not cover, or if you need a stable release rather than a Beta. Before you commit, verify three things yourself: that your Python version is 3.11 or 3.12 and not 3.14 or later, that your GPU path is one the project actually supports rather than one you assume it does, and that the extension you depend on is not the LLM prompting feature the README still lists as unimplemented. The last push to the repository was on 2026-09-09.

Frequently asked questions

Is SwarmUI free?

Yes. The README states that SwarmUI is 100 percent free and open source forever, and the repository is licensed under MIT. The project asks for donations on Patreon but does not paywall access.

What is SwarmUI used for?

It is a web user interface for AI image generation, AI video models and some AI audio models. The README lists support for image models such as Stable Diffusion and Flux, video models such as Wan and LTX-2, and audio models such as ACE-Step.

Which is better, ComfyUI or SwarmUI?

The README frames SwarmUI as a UI that contains a Comfy Workflow tab giving the unrestricted raw graph, plus a Generate tab that generates workflows for you. ComfyUI's graph is the interface itself, so SwarmUI adds a convenience layer on top rather than replacing the graph.

How do I install SwarmUI?

On Windows you download install-windows.bat from the 0.9.8-Beta release, store it outside Program Files and run it, running it a second time if it closes early. On Linux you download install-linux.sh, make it executable and run it, after installing git and python3 with Python 3.11 or 3.12.

Is SwarmUI good?

The README calls the project Almost-Release and recommends it as an ideal UI for most users, beginners and pros alike, while listing LLM-assisted prompting as not yet natively implemented. The Windows installer is documented as sometimes needing a second run.

Official sources

  1. Issues
  2. License: MIT
  3. mcmonkeyprojects/SwarmUI on GitHub
  4. README
  5. Releases
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/mcmonkeyprojects-swarmui.svg)](https://hysenlabs.com/projects/mcmonkeyprojects-swarmui)