Z-Image-Edit

Z-Image-Edit is a professional AI-powered image editing and generation platform developed by Alibaba Tongyi, featuring three model variants (Z-Image-Edit, Z-Image-Turbo, Z-Image-Base) for instruction-based editing, ultra-fast generation, and foundational development.

What is Z-Image-Edit?

Z-Image-Edit is an AI image editing and generation platform developed by Alibaba Tongyi. It takes text prompts and optional input images for editing, and produces output images in JPG format. The platform runs on a web interface and offers three model variants: Z-Image-Edit for precise editing via natural language instructions, Z-Image-Turbo for sub-second image generation, and Z-Image-Base for open-source foundational development. The underlying architecture is a Single-Stream Diffusion Transformer (S3-DiT) with 6 billion parameters.

Key Features

• Bilingual instruction support — Accepts editing commands in both Chinese and English for precise image modifications.
• Ultra-fast generation — Z-Image-Turbo variant completes image generation in under a second using only 8 inference steps, compatible with consumer GPUs with 16 GB VRAM.
• Single-Stream DiT architecture — Powered by a 6B-parameter S3-DiT model for efficient and high-quality image synthesis.
• Instruction-based image editing — Edit images by describing desired changes in natural language; supports style transfers, object replacements, and more.
• Photorealistic output — Generates high-quality photorealistic images, as demonstrated by strong Elo evaluation scores.
• Bilingual text rendering — Accurately renders Chinese and English text within generated images, useful for multilingual content.
• Credit-based usage — Each generation costs 5 credits; new users receive free credits upon signing in.

Who is it for?

• Digital artists — Quickly generate or edit concept art and illustrations using natural language commands.
• Advertising designers — Create ad materials with bilingual text support and rapid iterations.
• Game art designers — Generate characters, environments, and assets at 10x speed, with easy post-editing.
• Film post-production teams — Use instruction-based editing for VFX and scene adjustments.
• Product designers — Rapidly prototype product visuals and iterate without pressure.

What can you do with Z-Image-Edit?

• Text-to-image generation — Describe a scene or object in natural language and get a photorealistic image in seconds.
• Image-to-image editing — Upload an existing image and provide editing instructions (e.g., "change the background to a beach at sunset") for precise modifications.
• Bilingual text overlay — Generate images containing specific Chinese or English text, such as signs, labels, or captions.
• Style transfer — Apply artistic styles or transform the mood of an image through simple prompts.

How does Z-Image-Edit work?

1. Choose a model variant (Z-Image-Edit, Z-Image-Turbo, or Z-Image-Base) based on your task.
2. Input instructions — Type natural language commands in Chinese or English, optionally upload a reference image.
3. Generate — The AI processes your request and produces the resulting image.
4. Download and share — Export the high-quality JPG image for use in your projects.

Pricing

Z-Image-Edit uses a freemium model. New users receive free credits upon signing in; each generation costs 5 credits. Additional credits can be purchased from the pricing page.