.jpg)
Learn how to install ComfyUI and what are the requirements for ComfyUI
System Requirements for ComfyUI
Running ComfyUI requires a machine that can handle AI image generation workloads. Here’s what you need:
Minimum Requirements
- GPU: NVIDIA GPU with CUDA support (at least 6GB VRAM, e.g. GTX 1660 or RTX 2060).
- CPU: Modern quad-core processor.
- RAM: 8GB system memory.
- Storage: 10–20GB free space for installation, models, and checkpoints.
- OS: Windows 10/11, macOS (M-series Apple Silicon supported via Metal), or Linux.
Recommended Requirements (for smooth experience)
- GPU: RTX 3060 (12GB VRAM) or higher — more VRAM = faster generations, larger images.
- RAM: 16–32GB.
- Storage: SSD with 50–100GB free (models and workflows can add up fast).
- OS: Linux or Windows 11 (Linux gives slightly better performance).
💡 Note: You can run ComfyUI without a GPU, but generations will be very slow (minutes per image).
How to Install ComfyUI with Promptus
If all that feels like a lot of setup, there’s a much easier path: Promptus.
- Promptus has always supported ComfyUI — in fact, the best version of ComfyUI is packaged and optimized by Promptus.
- It’s free to use for paid Promptus users (subscription).
- Or, if you prefer ownership, you can buy the Promptus application once for $49 — and it includes ComfyUI out of the box, pre-configured and ready to run.
Even better: inside Promptus, you don’t just get ComfyUI. You can also use Google’s new Gemini 2.5 Flash Image (Nano Banana) for reliable image generation, editing, and fusion. That means you get both flexibility (ComfyUI) and next-gen consistency (Nano Banana) in one platform.
👉 Get started here: Promptus.ai
How to Install ComfyUI (Not using Promptus) ⚙️
Step 1: Install Dependencies
- Install Python 3.10 or 3.11.
- Install Git.
- (Windows users) Install CUDA Toolkit if using an NVIDIA GPU.
Step 2: Clone ComfyUI
Open a terminal (or PowerShell) and run:
git clone https://github.com/comfyanonymous/ComfyUI.git
cd ComfyUI
Step 3: Install Requirements
pip install -r requirements.txt
Step 4: Run ComfyUI
python main.py
ComfyUI will launch in your browser at http://127.0.0.1:8188/
.
Common Issues ⚡
- CUDA errors: Update your GPU drivers and ensure your CUDA toolkit matches your PyTorch version.
- Out of memory (OOM): Lower resolution or batch size; upgrade GPU if possible.
- Slow generation: If you don’t have a GPU, expect slow performance. Consider using a cloud provider.
Conclusion
So, how do you install ComfyUI?
- If you want the easiest way with extra features, Promptus is the fastest route — optimized ComfyUI plus access to Nano Banana, all in one app.
- If you want to manually install, it will take multiple attempts and some time, you can install it manually using Git + Python, provided you have a GPU that meets requirements.