Complete Guide to Running Z-Image Locally with ComfyUI (Plus an Easy Online Alternative)
On November 27th, Alibaba’s Tongyi Lab officially released the Z-Image image generation model, making it available not only for cloud inference but also for local deployment on consumer GPUs. Thanks to its lightweight 6B parameter design, the model runs smoothly on GPUs with 16GB VRAM and can even operate—under reduced settings—on devices with 8GB VRAM.
ComfyUI has also published an official workflow template for Z-Image Turbo, making the local installation process remarkably simple. This article provides a complete, beginner-friendly guide to setting up Z-Image locally using ComfyUI.
For users who prefer not to install anything or lack sufficient GPU resources, we also introduce the popular online alternative Z-Image-Edit.com (https://z-image-edit.com), a browser-based tool where you can generate images instantly with the Z-Image Turbo model—no setup required.
Recommended Hardware
- Best experience: RTX 4060 / 5060 / 4070 series with 16GB VRAM
- Minimum usable requirement: ~8GB VRAM
- OS support: Windows / Linux
Z-Image’s efficiency makes it far more accessible than heavier models such as FLUX or Stable Diffusion XL.
Step 1: Download and Install ComfyUI Desktop
Visit the official ComfyUI download page and install the desktop application.
Once installed, launch the app—no advanced configuration is needed.
Step 2: Download the Required Z-Image Model Files
Z-Image Turbo requires three core components:
- diffusion_model
- text_encoder
- vae
You can download them directly from:
https://huggingface.co/Comfy-Org/z_image_turbo/tree/main/split_files
After downloading, place each file into the appropriate model folder inside your ComfyUI installation directory (e.g., ComfyUI/models/...).
Step 3: Download the Official Workflow Template
ComfyUI provides a ready-to-use workflow for Z-Image Turbo:
https://github.com/Comfy-Org/workflow_templates/blob/main/templates/image_z_image_turbo.json
Save this file into: ComfyUI/user/default/workflows
Next time you open ComfyUI, the workflow will appear in your workflow list.
Step 4: Load the Workflow and Start Generating Images
Open ComfyUI and load the template you just downloaded.
Then:
- Enter your prompt
- Optionally adjust image width and height
- Click the Run button in the top-right corner
The workflow is preconfigured, meaning you can start generating images immediately without modifying nodes or settings.
Don’t Want Local Setup? Try Z-Image-Edit.com (No GPU Needed)
While local deployment is powerful, many creators face limitations:
- No dedicated GPU
- Insufficient VRAM
- Unwilling to manage installation
- Want to create images on multiple devices
For these users, an online tool becomes far more convenient.
Z-Image-Edit is a highly recommended free online service that has already integrated the Z-Image Turbo model.
Compared to local deployment, it's ideal for quick experimentation, lightweight devices, or users who simply want to focus on creation.
Final Thoughts
Whether you run it locally through ComfyUI or use it online via Z-Image-Edit.com, Z-Image Turbo provides an impressive balance of speed, quality, and accessibility. Local deployment offers full control and workflow customization, while the online tool offers effortless usability.
For creators, designers, and hobbyists, Z-Image is shaping up to be one of the most practical and efficient open-source image generation models available today.
