Skip to content
Follow me on X
•ComfyUI Wiki
NewsAlibaba Tongyi Lab Releases Z-Image-Turbo - Efficient 6B Parameter Image Generation Model

Alibaba Tongyi Lab Releases Z-Image-Turbo - Efficient 6B Parameter Image Generation Model

11/27/2025

Alibaba Tongyi Lab Releases Z-Image-Turbo - Efficient 6B Parameter Image Generation Model

On November 27, 2025, Alibaba Tongyi Lab officially released Z-Image-Turbo, a next-generation efficient image generation model. With only 6B parameters, this model achieves performance comparable to closed-source flagship models with 20B+ parameters, excelling particularly at generating high-fidelity, photorealistic portraits.

Key Features

Efficient Parameter Architecture

Z-Image-Turbo uses a Scalable Single-Stream DiT (S3-DiT) architecture that concatenates text, visual semantic tokens, and image VAE tokens at the sequence level as a unified input stream, maximizing parameter efficiency. This design allows the model to maintain high-quality output while significantly reducing computational resource requirements.

Consumer GPU Friendly

This is one of Z-Image-Turbo’s most attractive features:

  • VRAM usage strictly controlled to under 16GB
  • Generates 1024×1024 resolution images in just 2.3 seconds on RTX 4090
  • Generates 2K images in only 4.8 seconds on NVIDIA RTX Pro 6000 Blackwell
  • Produces high-quality images in just 8 sampling steps
  • Can even run on RTX 3060 6GB version

Excellent Text Rendering

Z-Image-Turbo excels at text rendering in both Chinese and English, capable of:

  • Accurately rendering complex Chinese and English text
  • Maintaining facial realism and visual aesthetics
  • Achieving results comparable to top closed-source models

Photorealistic Portraits

The model particularly excels at generating:

  • High-fidelity character portraits
  • Natural skin textures and lighting effects
  • Cinematic lighting and composition
  • Various portrait photography styles

Using in ComfyUI

Z-Image-Turbo is now supported in ComfyUI. You need to:

  1. Update ComfyUI to the latest version
  2. Download the required model files
  3. Import the workflow to start using

Model Files

You need to download the following model files and place them in the corresponding directories:

Text Encoders

Diffusion Models

VAE

File Storage Location

📂 ComfyUI/
├── 📂 models/
│   ├── 📂 text_encoders/
│   │      └── qwen_3_4b.safetensors
│   ├── 📂 diffusion_models/
│   │      └── z_image_turbo_bf16.safetensors
│   └── 📂 vae/
│          └── ae.safetensors

ComfyUI Workflow

You can download the official workflow template from the following link:

Future Plans

The research team plans to release the full, non-distilled foundation model, which will enable community-driven fine-tuning, custom workflows, and broader open-source ecosystem development.

Related Links