Download Studio, run your first Pipe, and borrow a cloud GPU whenever you need more VRAM.
Browse Pipes and pick one job to start with.
Download the .tea and open it in Studio.
Create inside a Project so outputs and settings stay together.
One click to the latest stable installers.
If you want a fast, disposable setup (or more VRAM than your local machine), deploy the Sweet Tea Image template on RunPod or Vast.ai.
Simple concepts that make capability feel like a workspace.
Sync to your Desk (Plus)
Verified-only browsing toggle (Plus)
Premium pipes (Plus-only downloads)
Plus friction relief
Plus adds extra Studio features like dependency management, one-click sync to your Desk, verified-only browsing, and premium pipes.
Installer
118.6 MB
Windows may show SmartScreen warnings for new binaries. If that happens: click More info then Run anyway.
Installs and updates preserve your local library and settings (stored under your user profile).
1. Create a GPU Pod and set the image to the tag below.
2. Expose ports 3000, 8000, 7778 (optional 8188).
3. (Recommended) Attach a persistent volume and set PERSISTENT_MOUNT_DIR=/workspace.
1. Create an instance with the image tag below.
2. Expose ports 3000, 8000, 7778 (optional 8188).
3. (Recommended) Use a persistent mount and set PERSISTENT_MOUNT_DIR=/data.
Docker image
jkotieno/comfyui-sweet-tea-cu128:2.0.0-sage3Optional: pin by digest
jkotieno/comfyui-sweet-tea-cu128@sha256:d016ecafe35e1043ecc2d88a8e79928c0799534f8fbc8534efb1db839ca151b8Expose ports
3000 (recommended): Nginx frontdoor (web proxy)
8000: Sweet Tea Studio UI/API
7778: Code Server (IDE + terminal)
8188: ComfyUI direct (optional)
Template source
Deployment image:Docker Hub
Security note: ComfyUI is typically unauthenticated by default. Treat the URLs as public and use platform-level access controls.