Z-Image-GGUF

This is a GGUF quantized version of Z-Image.

unsloth/Z-Image-GGUF uses Unsloth Dynamic 2.0 methodology for SOTA performance.

• Important layers are upcasted to higher precision.
• Uses tooling from ComfyUI-GGUF by city96.

https://github.com/unslothai/unsloth/ https://discord.gg/unsloth https://docs.unsloth.ai/basics/unsloth-dynamic-2.0-ggufs

⚡️- Image#

An Efficient Image Generation Foundation Model with Single-Stream Diffusion Transformer

Official Site  GitHub  Hugging Face  ModelScope Model  ModelScope Space  https://arxiv.org/abs/2511.22699

Welcome to the official repository for the Z-Image（造相）project!

🎨 Z-Image#

Image: Teaser Image: asethetic Image: diverse Image: negative

Z-Image is the foundation model of the ⚡️- Image family, engineered for good quality, robust generative diversity, broad stylistic coverage, and precise prompt adherence. While Z-Image-Turbo is built for speed, Z-Image is a full-capacity, undistilled transformer designed to be the backbone for creators, researchers, and developers who require the highest level of creative freedom.

Image: z-image

🌟 Key Features#

• Undistilled Foundation: As a non-distilled base model, Z-Image preserves the complete training signal. It supports full Classifier-Free Guidance (CFG), providing the precision required for complex prompt engineering and professional workflows.
• Aesthetic Versatility: Z-Image masters a vast spectrum of visual languages—from hyper-realistic photography and cinematic digital art to intricate anime and stylized illustrations. It is the ideal engine for scenarios requiring rich, multi-dimensional expression.
• Enhanced Output Diversity: Built for exploration, Z-Image delivers significantly higher variability in composition, facial identity, and lighting across different seeds, ensuring that multi-person scenes remain distinct and dynamic.
• Built for Development: The ideal starting point for the community. Its non-distilled nature makes it a good base for LoRA training, structural conditioning (ControlNet) and semantic conditioning.
• Robust Negative Control: Responds with high fidelity to negative prompting, allowing users to reliably suppress artifacts and adjust compositions.

🆚 Z-Image vs Z-Image-Turbo#

🚀 Quick Start#

Installation & Download#

Install the latest version of diffusers:

pip install git+https://github.com/huggingface/diffusers

Download the model:

pip install -U huggingface_hub
HF_XET_HIGH_PERFORMANCE=1 hf download Tongyi-MAI/Z-Image

Recommended Parameters#

• Resolution: 512×512 to 2048×2048 (total pixel area, any aspect ratio)
• Guidance scale: 3.0 – 5.0
• Inference steps: 28 – 50

Usage Example#

import torch
from diffusers import ZImagePipeline

# Load the pipeline
pipe = ZImagePipeline.from_pretrained(
    "Tongyi-MAI/Z-Image",
    torch_dtype=torch.bfloat16,
    low_cpu_mem_usage=False,
)
pipe.to("cuda")

# Generate image
prompt = "两名年轻亚裔女性紧密站在一起，背景为朴素的灰色纹理墙面，可能是室内地毯地面。左侧女性留着长卷发，身穿藏青色毛衣，左袖有奶油色褶皱装饰，内搭白色立领衬衫，下身白色裤子；佩戴小巧金色耳钉，双臂交叉于背后。右侧女性留直肩长发，身穿奶油色卫衣，胸前印有“Tun the tables”字样，下方为“New ideas”，搭配白色裤子；佩戴银色小环耳环，双臂交叉于胸前。两人均面带微笑直视镜头。照片，自然光照明，柔和阴影，以藏青、奶油白为主的中性色调，休闲时尚摄影，中等景深，面部和上半身对焦清晰，姿态放松，表情友好，室内环境，地毯地面，纯色背景。"
negative_prompt = "" # Optional, but would be powerful when you want to remove some unwanted content

image = pipe(
    prompt=prompt,
    negative_prompt=negative_prompt,
    height=1280,
    width=720,
    cfg_normalization=False,
    num_inference_steps=50,
    guidance_scale=4,
    generator=torch.Generator("cuda").manual_seed(42),
).images[0]

image.save("example.png")

📜 Citation#

If you find our work useful in your research, please consider citing:

@article{team2025zimage,
  title={Z-Image: An Efficient Image Generation Foundation Model with Single-Stream Diffusion Transformer},
  author={Z-Image Team},
  journal={arXiv preprint arXiv:2511.22699},
  year={2025}
}