Z-Image Turbo
An open-weight, high-speed text-to-image model from Alibaba’s Tongyi Lab, optimized for photorealism and bilingual text rendering with minimal inference steps.
Generate imageGet API keyWhat is Z-Image Turbo?
Z-Image Turbo is a 6-billion-parameter text-to-image model developed by Alibaba’s Tongyi Lab, released in November 2025. It generates photorealistic images in under 3 seconds using only 8 inference steps, supports bilingual text rendering in English and Chinese, and runs efficiently on consumer GPUs with 16GB VRAM.
Use Z-Image Turbo privately on Venice
On Venice, Z-Image Turbo runs with zero retention — your prompts are never stored or used for training. This open-weight model is available privately, letting you generate images without surveillance or data harvesting. Its speed and efficiency make it ideal for developers and creators who want fast, uncensored, and sovereign AI image generation.
What can Z-Image Turbo do?
- •Extremely fast generation — sub-second inference on H800 GPUs, under 3 seconds on consumer 16GB VRAM hardware.
- •Efficient 8-step NFE (Number of Function Evaluations) enables low-latency, high-throughput workflows.
- •Strong photorealistic image quality for a model of its size, especially in portraits and character generation.
- •Bilingual text rendering in both English and Chinese, a rare and valuable capability.
- •Open weights under Apache 2.0 license allow for full commercial use, modification, and local deployment.
- •Lower diversity in outputs compared to larger or more complex models, per official documentation.
- •No classifier-free guidance (CFG) support, limiting fine control over prompt adherence.
- •Fine-tunability is limited or not officially supported, reducing customization options.
- •May struggle with highly complex or layered prompts that work well on other diffusion models.
Sample outputs
Generated on Venice with our standard prompt suite — the same prompts we run through every model of this type, so you can judge it like-for-like.
A retro travel poster with the bold headline "VENICE" in large condensed serif type, sunset color palette, clean layout
Photorealistic close-up portrait of a weathered fisherman at golden hour, 85mm lens, shallow depth of field, natural skin texture
A small red cube balanced on top of a large glossy blue sphere, with a green cone to the right, plain light-grey studio background
Cozy watercolor illustration of a hillside village in autumn, warm tones, soft paper texture
How to use Z-Image Turbo via API
Venice exposes this model through the REST API. Queue a generation with z-image-turbo.
curl https://api.venice.ai/api/v1/image/generate \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "z-image-turbo",
"prompt": "A serene mountain lake at dawn, photorealistic"
}' --output image.pngSpecifications
Pricing
Flat per-image pricing on Venice: $0.01 per generation.
New Venice accounts include a free daily allowance and 500 welcome credits — no credit card required.
Z-Image Turbo vs alternatives
| Model | Max resolution | Strongest at | Open weights | Price (Venice) |
|---|---|---|---|---|
| Z-Image Turbo | — | Speed & photorealism | Yes | $0.01 / image |
| Chroma | — | Speed & simplicity | Yes | $0.01 / image |
| Grok Imagine High Quality (SOTA) | — | Aesthetic quality | No | from $0.06 / image |
| Flux 2 Max | — | Photoreal detail | No | $0.09 / image |
Fast, open, and efficient — ideal for developers and local deployment.
What is Z-Image Turbo good for?
- •Fast prototyping for developers and AI researchers testing image generation pipelines.
- •E-commerce content creation requiring accurate bilingual product visuals.
- •Local, privacy-preserving image generation on consumer hardware without cloud dependency.
- •API-based services needing low-cost, high-speed image generation with commercial rights.
- •Applications requiring open, auditable models without proprietary restrictions.
Prompting tips
- •Keep prompts concise and direct — Z-Image Turbo performs best with clear, structured inputs.
- •Use explicit language for text rendering: include phrases like 'text in English' or 'Chinese characters' when needed.
- •Avoid overly complex or nested descriptions; simpler prompts yield more reliable results.
- •Leverage its speed for rapid iteration — generate multiple variants quickly to find the best output.
Version history
Base model with higher quality and diversity.
CurrentDistilled for speed and efficiency.
Frequently asked questions
Z-Image Turbo is a 6-billion-parameter text-to-image model developed by Alibaba’s Tongyi Lab, released in November 2025. It generates photorealistic images in under 3 seconds using only 8 inference steps, supports bilingual text rendering in English and Chinese, and runs efficiently on consumer GPUs with 16GB VRAM.
On Venice, Z-Image Turbo costs $0.01 per image generation, with upscaling priced at $0.02 (2×) and $0.08 (4×). There are no subscription fees — you pay only for what you generate.
Yes. Z-Image Turbo is open weights under the Apache 2.0 license, meaning you can download, modify, and use it commercially at no cost. The model weights are freely available on Hugging Face and ModelScope.
No. Z-Image Turbo is a pure text-to-image model without built-in tool use, web search, or reasoning capabilities. It generates images directly from text prompts without external data retrieval.
Yes. It excels at rendering bilingual text in both English and Chinese within generated images, a feature many competing models struggle with. This makes it valuable for multilingual content creation.
Z-Image Turbo is faster, cheaper, and open, making it ideal for developers and cost-sensitive workflows. Flux 2 Max offers higher photorealistic fidelity and artistic control but at a higher price and without open weights. Your choice depends on whether you prioritize openness and speed or peak image quality.
Yes. It runs on consumer GPUs with 16GB VRAM and is designed for local deployment. The open weights and Apache 2.0 license allow full offline use, fine-tuning, and integration into custom workflows.
Related models
Run Z-Image Turbo privately.
No prompt logging. No data used for training. Free to start — no credit card.
