Z-Image Turbo

Overview

Z-Image Turbo is the official distilled model from Tongyi-MAI, the research division behind the Z-Image family. With 6 billion parameters and an optimized inference pipeline, it delivers high-quality photorealistic images in just 8 steps, making it one of the fastest foundation models available. It serves as the base for many popular community fine-tunes.

Key Features

• Ultra-Fast Generation: Distilled architecture produces detailed images in only 8 inference steps
• Photorealistic Quality: Strong baseline for realistic imagery with natural proportions and lighting
• Bilingual Text Rendering: Capable of rendering both English and Chinese text within images
• Strong Instruction Adherence: Follows complex, multi-element prompts with high fidelity
• Foundation Model: The base checkpoint behind many community fine-tunes in the Z-Image ecosystem

Best For

• General-purpose photorealistic image generation
• Portraits and people photography
• Landscape and architectural scenes
• Commercial and product imagery
• Quick iterations and concept exploration

Prompt Tips

• Z-Image Turbo works well with natural, conversational prompts. Describe your scene as you would explain it to a photographer.
• The model responds accurately to detailed prompts, so the more specific you are, the closer the result matches your vision
• For photorealistic results, mention camera and lens details (e.g., "shot on 85mm lens", "shallow depth of field")
• It handles complex scenes with multiple elements well. Don't hesitate to describe a full composition.