ComfyUI: Run AI Workflows Locally, Fast, and Free
A modular, graph-based Stable Diffusion engine—runs completely offline, adaptable, and open source.
Download ComfyUI
| Platform | Requirements | Install Method |
|---|---|---|
| Windows | NVIDIA GPU required | Download Installer |
| macOS | Apple Silicon (M1, M2, M3) | Download Installer |
| All Platforms | — | Install from GitHub |
Why ComfyUI
- Fully offline execution—local compute, minimal latency, no cloud fees. :contentReference[oaicite:1]{index=1}
- Visual pipeline builder—nodes and flowcharts make complex Stable Diffusion workflows accessible without code. :contentReference[oaicite:2]{index=2}
- Totally open source—completely free, community-extendable. :contentReference[oaicite:3]{index=3}
Installation Options
- Windows (NVIDIA): Click the installer link above.
- macOS (Apple Silicon): Use the macOS installer link.
- From GitHub: Fetch and build manually via the ComfyUI repo—great for power users. :contentReference[oaicite:4]{index=4}
Additional Resources
Install via GitHub (Manual & Advanced)
While installers are available for casual users, ComfyUI’s full flexibility is unlocked through GitHub.
Follow the steps below to install ComfyUI from source and enable custom workflows on your preferred platform.
🧰 System Requirements
- Python 3.10–3.12 recommended (3.13 partially supported)
- NVIDIA GPU with CUDA support for best performance (RTX 20xx+)
- Apple Silicon (M1/M2/M3) supported via PyTorch nightly
- Intel/AMD/DirectML/Ascend/MLU users can use CPU fallback (
--cpu) - Minimum 8GB RAM; 16GB+ preferred for complex workflows
Tip: If your GPU fails to initialize, try launching with --cpu or check your torch install.
📦 Install Steps (Terminal / PowerShell)
- Install Python 3.12 and
git - Open terminal, then run:
git clone https://github.com/comfyanonymous/ComfyUI.git cd ComfyUI python -m venv venv venv\Scripts\activate # On Linux/macOS: source venv/bin/activate pip install -r requirements.txt python main.py - ComfyUI will launch in your browser at
http://127.0.0.1:8188
🚀 Alternate Install Platforms
- Apple Silicon (M1/M2/M3): Use PyTorch nightly before pip install. Full install guide in README.
- Intel GPUs (OneAPI): Try the
xpu-nightlytorch build with environment variableCO