Z Image Base (LoRA)

Z Image Base (LoRA) Z-Image Base LoRA is the undistilled foundation of the Z-Image family, intended for customization and diverse prompt-led generation. The attached low-rank adaptation introduces a specialized aesthetic, subject, or production concept without replacing the underlying image model.

Use this foundation for research, fine-tuning, custom pipelines, and projects that value diversity over minimum latency. Include the adaptation’s learned concept naturally inside the brief so it interacts with the complete composition.

Highlights

• Z-Image Base LoRA Undistilled image foundation. Retains the full generative training signal for adaptation and research.
• Z-Image Base LoRA High output diversity. Explores a wider range of compositions and visual interpretations.
• Z-Image Base LoRA Fine-tuning readiness. Provides a suitable foundation for domain, style, or character adaptation.
• Z-Image Base LoRA Detailed text-to-image synthesis. Builds polished photographic and illustrative scenes from descriptive prompts. A LoRA adds a learned style, subject, or domain concept to this behavior.

Pricing

When to Use

Prompt Guide

For generation, state the intended result first, then add the subject or source treatment, progression, style, and delivery constraints. Keep one creative variable per phrase, use the documented field names for controls, and change one setting at a time when comparing results.

{
  "aspect_ratio": "21:9",
  "output_format": "png",
  "prompt": "Grandmother knitting by a window, an empty chair by her"
}

Technical Specs

Related Models

• alibaba/z-image/base — Compare this concrete local family route.
• alibaba/z-image/turbo — Compare this concrete local family route.
• alibaba/z-image/turbo/edit — Compare this concrete local family route.