prunaai/z-image-turbo-lora 🔢📝❓ → 🖼️

▶️ 38.2K runs 📅 Dec 2025 ⚙️ Cog 0.16.9

About

Lora version of Z-Image Turbo, which is a super fast text-to-image model of 6B parameters developed by Tongyi-MAI.

Example Output

Prompt:

"t3chnic4lly vibrant 1960s close-up of a woman sitting under a tree in a blue skit and white blouse, she has blonde wavy short hair and a smile with green eyes lake scene by a garden with flowers in the foreground 1960s styl;e film She's holding her hand out there is a small smooth frog in her palm, she's making eye contact with the toad."

Output

Performance Metrics

2.39s Prediction Time

2.40s Total Time

All Input Parameters

{
  "width": 1024,
  "height": 1024,
  "prompt": "t3chnic4lly vibrant 1960s close-up of a woman sitting under a tree in a blue skit and white blouse, she has blonde wavy short hair and a smile with green eyes lake scene by a garden with flowers in the foreground  1960s styl;e film   She's holding her hand out there is a small smooth frog in her palm, she's making eye contact with the toad.",
  "lora_scales": [
    1
  ],
  "lora_weights": [
    "https://huggingface.co/renderartist/Technically-Color-Z-Image-Turbo/resolve/main/Technically_Color_Z_Image_Turbo_v1_renderartist_2000.safetensors"
  ],
  "output_format": "jpg",
  "guidance_scale": 0,
  "output_quality": 80,
  "num_inference_steps": 8
}

Input Parameters

seed Type: integer: Random seed. Set for reproducible generation
width Type: integerDefault: 1024Range: 64 - 1440: Width of the generated image
height Type: integerDefault: 1024Range: 64 - 1440: Height of the generated image
prompt (required) Type: string: Text prompt for image generation
lora_scales Type: array: Scales for each LoRA as an array of floats. Must match the number of lora_weights. Defaults to 1.0 for each if not provided.
lora_weights Type: array: LoRA weights as an array of URLs. Supports .safetensors URLs, tar files, and zip files (e.g., ['https://example.com/lora1.safetensors', 'https://example.com/lora2.safetensors'])
output_format Default: jpg: Format of the output images
guidance_scale Type: numberDefault: 0Range: 0 - 20: Guidance scale. Should be 0 for Turbo models
output_quality Type: integerDefault: 80Range: 0 - 100: Quality when saving the output images, from 0 to 100. 100 is best quality, 0 is lowest quality. Not relevant for .png outputs
num_inference_steps Type: integerDefault: 8Range: 1 - 50: Number of inference steps. This actually results in (num_inference_steps - 1) DiT forwards

Output Schema

Output

Type: array • Items Type: string • Items Format: uri

Example Execution Logs

Using cached processed LoRA for: https://huggingface.co/renderartist/Technically-Color-Z-Image-Turbo/resolve/main/Technically_Color_Z_Image_Turbo_v1_renderartist_2000.safetensors
Loaded LoRA 1: https://huggingface.co/renderartist/Technically-Color-Z-Image-Turbo/resolve/main/Technically_Color_Z_Image_Turbo_v1_renderartist_2000.safetensors
Set adapters: ['lora_0'] with weights: [1.0]
  0%|          | 0/9 [00:00<?, ?it/s]
 11%|█         | 1/9 [00:00<00:01,  4.42it/s]
 22%|██▏       | 2/9 [00:00<00:01,  5.15it/s]
 33%|███▎      | 3/9 [00:00<00:01,  4.79it/s]
 44%|████▍     | 4/9 [00:00<00:01,  4.63it/s]
 56%|█████▌    | 5/9 [00:01<00:00,  4.55it/s]
 67%|██████▋   | 6/9 [00:01<00:00,  4.50it/s]
 78%|███████▊  | 7/9 [00:01<00:00,  4.48it/s]
 89%|████████▉ | 8/9 [00:01<00:00,  4.45it/s]
100%|██████████| 9/9 [00:01<00:00,  4.44it/s]
100%|██████████| 9/9 [00:01<00:00,  4.54it/s]

Version Details

Version ID: 197b2db2015aa366d2bc61a941758adf4c31ac66b18573f5c66dc388ab081ca2
Version Created: December 4, 2025

Run on Replicate →