XaiJu
AI Knowledge Central
AI Knowledge Central

patreon


Update September 2025: Fix Cuda Error Nvidia 50 Series ForgeUI

Hey there, Friends! 🔧
Having trouble getting Forge UI up and running for your Stable Diffusion setup? No worries—I’ve got your back! Below are all the links and commands you need to get it fixed and running smooth like butter. 🍞⚙️

💛 Supporter Perk (Paying Supporters Only)
I’ve made a one-click installer for RTX 5090 / CUDA 12.8 that does everything for you (clone → venv with Python 3.10 → Torch cu128 stable→nightly fallback → first-run setup), plus auto-logging and a couple extra guardrails. It’s available to my paying supporters here on Patreon. If you want the comfy “click once and chill” experience, that’s the way. 🙌
https://www.patreon.com/posts/forge-ui-fresh-137804391

You can find the Video here:
https://www.youtube.com/watch?v=PtGgjdw5koA

🔗 1) Download Python 3.10.x (64-bit)
Forge UI plays best with Python 3.10.11. Grab it from the official site:
https://www.python.org/downloads/release/python-31011/

During install: check “Install launcher for all users” and “Add python.exe to PATH”.

🐍 List all Python versions you have
py -0p

✅ Confirm Python 3.10 is available
py -3.10 -V

🔗 2) Get Git (if you haven’t already)
https://git-scm.com/

🧪 Check Git
git --version

📦 3) Clone the Forge UI Repository
(Choose a folder first, e.g. E:\AI)
git clone https://github.com/lllyasviel/stable-diffusion-webui-forge.git

cd stable-diffusion-webui-forge

🧰 4) Create and Activate a fresh Python 3.10 venv
py -3.10 -m venv venv
venv\Scripts\activate
(optional) upgrade pip:
python -m pip install -U pip

⚡ 5) Install PyTorch for CUDA 12.8 (cu128)
install cuda 12.8
python -m pip install --upgrade torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

🔎 Quick CUDA check (you want “CUDA avail: True”)
python -c "import sys, torch; print('Torch', torch.version, 'CUDA', getattr(torch.version,'cuda',None), 'CUDA avail:', torch.cuda.is_available()); sys.exit(0 if torch.cuda.is_available() else 1)"

📌 6) Pin Python & Torch behavior in webui-user.bat
Create a file named: webui-user.bat in the Forge folder with this content:

@echo off
set "PYTHON=%~dp0venv\Scripts\python.exe"
set "VENV_DIR=venv"
set "TORCH_COMMAND=pip install --upgrade torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

set "COMMANDLINE_ARGS="
call webui.bat

(That keeps Forge on Python 3.10 and CUDA 12.8 wheels.)

🚀 7) First Run (setup only), then Launch
Run setup (installs remaining deps and exits):
webui-user.bat

Open webui-user.bat and remove “--exit”, then run it again to launch the UI.

Default URL:

http://127.0.0.1:7860

🧯 Troubleshooting
• INCOMPATIBLE PYTHON VERSION (shows 3.13.x):
rmdir /s /q venv
py -3.10 -m venv venv
venv\Scripts\activate

• Torch installed but CUDA not available (CUDA avail: False):
Make sure you used the cu128 index. If stable fails, use the nightly cu128 line above.

• Windows CMD error “and was unexpected at this time.”
That’s a batch parser quirk. Use the plain commands above as written (no fancy blocks), or grab my one-click installer if you’re a supporter.

💥 That’s it! You’re now set up and ready to roll with Forge UI on your local machine.
If this helped, drop a like—or join as a supporter to get the one-click installer & extras. 🎬🖼️

Stay creative and break things (in a good way)! 👨‍💻💡
— Chris

Update September 2025: Fix Cuda Error Nvidia 50 Series ForgeUI

Comments

Just wanted to say thank you for posting that great video and also these links; I couldn't get SD to work in other tutorials and yours worked; thank you!

Commander Riker

I have been using this guide previous and it worked good with FORGE

LVNDSCAPE

Hi, i am not sure if this works but try to go into your forge folder open a cmd and go to: Stable Diffusion WebUI Forge\venv\Scripts> there type "call activate" now (venv) should be before the folder path. There type: python -m pip install --upgrade pip setuptools wheel # Install PyTorch built for CUDA 12.8 (cu128) pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128 # Then install Forge’s requirements (let Forge/requirements.txt handle the rest) pip install -r requirements_versions.txt # or requirements.txt in your Forge folder I hope this will help

Chris Wenzl

Hi Chris, thanks! I did all this and yet I still get the error, as if cuda 12.8 is not recognized. But I do have it for a while (running comfy on my 5090). Any idea why? Danke!

BigBullAI Jim


More Creators