Z-Image-Turbo-GGUF by unsloth

Z-Image-Turbo-GGUF by unsloth

Unsloth's GGUF files of Z-Image-Turbo, a fast distilled text-to-image model that the authors say fits on a 16 GB consumer GPU. For people who generate images locally with GGUF-capable tools.

Made by unsloth on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: This repository holds Unsloth's GGUF quantized files of Z-Image-Turbo, a distilled text-to-image model from Tongyi-MAI. The files use Unsloth Dynamic 2.0, which raises the precision of important layers. The Hugging Face record lists 6.2B parameters and about 3.1 GB at 4-bit.

Who it suits: People who want to generate images from text on their own computer, and who use tools that can read GGUF image models.

Industry | Who uses it | How they use it | Example
Design | Designers drafting visuals | Generate images from a prompt locally | Make a concept image for a poster
Hobby art | Local AI users | Run a fast image model on a home GPU | Try many prompts in a short time
Research | People comparing quant sizes | Test GGUF versions of an image model | Compare outputs of two quant sizes

You can skip it if you want the original full-precision weights, since this repository holds quantized files only.

Where to use it:

• Tools that read GGUF image models, such as ComfyUI with GGUF support
• Consumer GPUs, since the authors say the model fits within 16 GB of VRAM
• Windows, macOS or Linux with Python, to download the files

Use case | Who | What they get back
Text-to-image | Designers | A picture from a written prompt
Fast iteration | Hobbyists | Quick results, since the model runs in 8 steps
Small-GPU generation | Home users | Image generation on a consumer card

How to set it up: You need a tool that can load GGUF image models and a GPU with enough memory for the file you pick. The part of the card used here does not show the run steps, so check the model page.

• Python and pip
• A GGUF-capable image tool
• A GPU that fits the quant you choose

1. Download a GGUF file.
2. Place it where your tool expects model files.
3. Follow the tool's workflow for Z-Image-Turbo.

pip install -U huggingface_hub
hf download unsloth/Z-Image-Turbo-GGUF

How to use it: Load the GGUF file in your image tool and type a text prompt. The authors describe the Turbo version as distilled so that it needs only 8 function evaluations.

Example: You write a prompt for a street scene at dusk and generate with the Turbo model. What comes back is one image in a few seconds on a capable GPU. The card gives no timing for this exact setup, and the examples shown there are pictures.

For newcomers and experts: If you are new, start with one prompt and the default settings in your tool before trying other quant files. Smaller files use less memory.

For experts, Z-Image has a single-stream diffusion transformer design with about 6B parameters. Turbo is the distilled version, and Base and Edit are the other variants. Quantizing trades file size for fidelity, and Unsloth raises the precision of key layers to limit the loss.

Text prompt
Z-Image-Turbo in GGUF, 8 steps
Generated image

Good to know:

• Size: about 6.2B parameters, around 3.1 GB at 4-bit by estimate
• Variants: Z-Image has Turbo, Base and Edit versions, and this repository covers Turbo
• Language: English listed
• License: Apache 2.0
• Cost to run: no API key or paid service named
• Data it reaches: downloads files from Hugging Face, nothing else is named
• Not for: image editing, since that is a separate Edit variant

On Hugging Face, 13 November 2026

• Task: text to image
• Library: ggml
• Parameters: 6.2B
• Weights: about 3.1 GB at 4-bit (estimate)
• Formats: GGUF
• Downloads in the last 30 days: 910,633
• Likes: 300
• License: apache-2.0
• Languages: en
• Based on: Tongyi-MAI/Z-Image-Turbo
• Tags: ggml, gguf, quantized, unsloth, text-to-image

Links

• Model: https://huggingface.co/unsloth/Z-Image-Turbo-GGUF
• Files: https://huggingface.co/unsloth/Z-Image-Turbo-GGUF/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download unsloth/Z-Image-Turbo-GGUF

Then follow the usage in its model card. Nothing in this kit runs by itself.