Instructions to use felipedpm/z-image-turbo-GGUF-confyui with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use felipedpm/z-image-turbo-GGUF-confyui with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL
# Run inference directly in the terminal:
llama cli -hf felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL
# Run inference directly in the terminal:
llama cli -hf felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL
# Run inference directly in the terminal:
./llama-cli -hf felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL
# Run inference directly in the terminal:
./build/bin/llama-cli -hf felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL

Use Docker

docker model run hf.co/felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL

LM Studio
Jan
Ollama
How to use felipedpm/z-image-turbo-GGUF-confyui with Ollama:
```
ollama run hf.co/felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL
```

Unsloth Studio

How to use felipedpm/z-image-turbo-GGUF-confyui with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for felipedpm/z-image-turbo-GGUF-confyui to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for felipedpm/z-image-turbo-GGUF-confyui to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for felipedpm/z-image-turbo-GGUF-confyui to start chatting

How to use felipedpm/z-image-turbo-GGUF-confyui with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

OpenClaw new

How to use felipedpm/z-image-turbo-GGUF-confyui with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Docker Model Runner
How to use felipedpm/z-image-turbo-GGUF-confyui with Docker Model Runner:
```
docker model run hf.co/felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL
```

Lemonade

How to use felipedpm/z-image-turbo-GGUF-confyui with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL

Run and chat with the model

lemonade run user.z-image-turbo-GGUF-confyui-UD-Q5_K_XL

List all available models

lemonade list

Hermes Agent

How to use felipedpm/z-image-turbo-GGUF-confyui with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default felipedpm/z-image-turbo-GGUF-confyui:UD-Q5_K_XL

Run Hermes

hermes

Atomic Chat

z-image-turbo (GGUF Version) for ComfyUI

This repository contains the GGUF quantized weights for the z-image-turbo model, optimized to run in environments with limited VRAM resources (though still demanding) using ComfyUI.

The goal of this upload is to enable the execution of this pipeline by leveraging the efficiency of the GGUF format for both the UNET and the Text Encoder (Qwen). Feel free to download and use just the workflow (json) in the models tab and versions!

Example Comparison

📂 Files and Structure

The models are organized within the repository folders as follows:

UNET: models/unet/z_image_turbo-Q8_0.gguf
- Q8 quantized version of the main diffusion model.
Text Encoder: models/text_encoders/Qwen3-4B-UD-Q5_K_XL.gguf
- Qwen3 4B LLM quantized in Q5, used for prompt processing.
VAE: models/vae/ae.safetensors
- Standard Variational Autoencoder for decoding the image.
Feel free to download and use just the workflow!

⚙️ Installation in ComfyUI

To use these models, you will need custom Nodes that support GGUF loading (such as City96's ComfyUI-GGUF or similar).

Download the files:
- Move the .gguf file from the unet folder to: ComfyUI/models/unet/
- Move the .gguf file from the text_encoders folder to: ComfyUI/models/clip/ (or text_encoders depending on your node loader).
- Move the .safetensors file from the vae folder to: ComfyUI/models/vae/
Recommended Nodes:
- Use UnetLoaderGGUF to load z_image_turbo-Q8_0.gguf.
- Use a GGUF-compatible CLIP/Text Encoder Loader to load Qwen3-4B.

💻 Hardware Requirements and Performance

⚠️ Warning: This is a heavy workflow.

Even with GGUF quantization, the model requires considerable hardware due to the size of the text encoder and the UNET.

Minimum GPU: 12GB VRAM (NVIDIA RTX 3060/4070 or higher).
System RAM: 32GB Recommended (System may need to offload data to RAM).

Generation Time

On a GPU with 12GB VRAM, the estimated generation time per image ranges between:

15 to 30 seconds. making considerably fast!

Model Information Check out the original model card Z-Image Turbo for detailed information about the model.

🔗 Useful Links

Downloads last month: 2,076

GGUF

Model size

4B params

Architecture

qwen3

Hardware compatibility

5-bit

8-bit