We have been following the Automatic1111 and Native Forge repositories for a long time, but there haven't been any updates adding support for new diffusion-based models since late 2024. You might wonder why no one is building support for these WebUIs anymore. The developers are busy in building something different.
Now, that is where Forge Neo (from different developer-Haoming02) comes in, an upgraded version of ForgeUI, built on the Gradio library. It also provides you to use symlinks, so you can point your diffusion models to a common directory shared with ComfyUI.
 |
| Forge Neo Dashboard |
Features
Video Generation -Wan 2.2 with high/low noise(txt2img, img2img, txt2vid, img2vid) plux built-in Video Player.
Image generation - Nunchaku (flux-dev, flux-krea), Flux-Kontext (img2img, inpaint), Chroma (t2I, Img2img), Qwen-Image
Image editing - Reactor, Qwen-Image-Edit, Flux-kontext (img2img, inpaint)
Optimized code - For faster inference
API support- Also has API integration functionalities like older one.
Installation
If you are a ForgeUI or an Automatic1111 user, its recommended and better to choose the new folder location for Forge Neo installation otherwise you will encounter with multiple errors. You can of course upgrade your older WebUI to the Neo variant but you need to be techy enough and know what your are doing. The developer has just removed almost half of the codes from original one.
Right now we are showing the recommended way on windows platform. Linux users need to follow the same procedure(in place of .bat, use .sh files instead) as on basic Forge installation. So we chose new folder location E Drive. Here, open command prompt and follow these installation commands:
Neo WEBUI Installation
1. Install git from its official page.
Note-If you already installed git and python then setup these isn't required. Just Skip step 1 and 2.
2. Install & Setup Python
(a) Standard Method with pip management-
Install python 3.11.9 and check the Add Python to PATH option.
(b) Alternative Method with uv setup-
Install uv tool for faster project manager.
Note-you will be thinking why "uv". Its a mordern feature over traditional "pip" that provides you high-speed package and project management.
Set up venv (virtual environment)cd sd-webui-forge-neo
uv venv venv --python 3.11 --seed
Add the --uv flag to webui-user.bat
3. Clone the Forge Neo Repository:git clone https://github.com/Haoming02/sd-webui-forge-classic sd-webui-forge-neo --branch neo
4. Finally, open the WebUI by clicking on webui-user.bat file. At the initial launch, it will download and automatically install all the required files (Pytorch, cuda, sage attention etc). So, be patient and wait for it. Once finished installation process, the WebUI will trigger the browser to open on port:
http://127.0.0.1:7860
5. Enjoy image and video Generation.
Generated image using Flux Dev model.
Prompt: a photo of an abandoned studio store with neon sign that says "FORGE NEO"
Steps-24
Sampler-euler
CFG- 5
SageAttention 2 Installation (Optional for RTX 5000 users)
People using RTX 5000 series cards may require to install Sage Attention manually. It provides you faster inference. Its based on Quantization so you will face little image/video quality degradation but its minimal. When you use this feature, both positive & negative prompting needed otherwise this cause an error.
1. Open your command prompt inside Webui installation folder. Now use following command to move into "sd-webui-forge-neo" folder :cd sd-webui-forge-neo
2. Activate your virtual environment:venv\scripts\activate
3. Create a folder as :mkdir repo
cd repo
4. Now, clone the sage attention repository:git clone https://github.com/thu-ml/SageAttention
If your CUDA version mismatches, just follow the SageAttention repository and setup .whl files.
Move into sage attention folder.cd SageAttention
5. Install the dependencies:-
for pip users, use this commandpip install -e . --no-build-isolation
For "uv" users as expained above, use this command: uv pip install -e . --no-build-isolation
Older Pytorch Installation (Optional)
1. If you want then install older pytorch in this way. Inside WebUI folder, search for "webui-user.bat" file. Open using any editor(Notepad++/ VS code etc). We are using VS Code.
2. Edit and add these new commands with parameters :
set TORCH_COMMAND=pip install torch==2.1.2 torchvision==0.16.2 --extra-index-url https://download.pytorch.org/whl/cu121
3. Save and close the file.
Installing FFmpeg (for windows users)
1. Download FFmpeg (ffmpeg-git-essentials.7z). This will help you in exporting your generated AI videos.
2. Extract file using any ZIP extration tool (Wainrar/7Zip etc). You will get a folder named something like:
"ffmpeg-7.1-full_build" or "ffmpeg-2025-essentials" as folder. It may be other name if you have new version.
3. Setup the path for environment variables. To do this, rename the folder for convenience as "ffmpeg" as folder name. Move the whole ffmpeg folder to any drive.
We chosen C Drive. If you are selecting different drive just change the drive name in command accordingly provided below:
C:\ffmpeg
So the full path of the executable will become:
C:\ffmpeg\bin\ffmpeg.exe
Now press Windows Key, search for "environment variables". Then, edit the system environment variables.
Edit the system environment variables,Click Environment Variables button.
Under System variables, select Path, then click Edit option.
Click New and paste this path:C:\ffmpeg\bin
Now, select OK everywhere to save your configurations.
4. At last to verify installation, type into command prompt:ffmpeg -version
And this will prints version "FFmpeg installed successfully". If you encounter errors like "'ffmpeg' is not recognized", this means you have setup environment variables wrongly.
Installing FFmpeg (for Linux users)
1. Open the linux terminal and update using following command:sudo apt update
2. Next, install FFmpeg:sudo apt install ffmpeg
3. After installation, verify it by typing:ffmpeg
