How fast is Z Image Turbo in real projects?
Z Image Turbo runs an optimized 8-step diffusion pipeline, so most scenes finish in 2–4 seconds on the hosted cluster and under 10 seconds when you self-host with a mid-range GPU. Because Z Image Turbo keeps prompts warm in memory, you can audition multiple compositions without breaking your flow.
Do I need special hardware or can I just run Z Image Turbo online?
You can open Z Image Turbo in the browser and let our managed GPUs do the heavy lifting, or download the Apache-licensed weights to ComfyUI, RunPod, or bare metal. The same Z Image Turbo model files power both flows, so you can switch anytime.
Can I use Z Image Turbo images for clients and commercial work?
Yes. Z Image Turbo ships under permissive terms, so every render you produce belongs to you. Agencies run Z Image Turbo for marketing decks, merch, and product shots without hidden royalties.
Does Z Image Turbo support bilingual prompts and Chinese typography?
Z Image Turbo was distilled with bilingual text encoders, so English, Chinese, and mixed prompts stay coherent. Even when you spell out Chinese slogans, Z Image Turbo keeps the strokes clean instead of producing garbled glyphs.
Is Z Image Turbo really free, and how do credits work?
Self-hosted Z Image Turbo is 100% free—download the checkpoints, deploy once, and generate forever. Our hosted Z Image Turbo tiers simply add convenience: subscribe for bundled credits or top up pay-as-you-go when you only need a few bursts.
How do I deploy or integrate Z Image Turbo into a pipeline?
Export the Z Image Turbo GGUF or safetensors weights, drop them into ComfyUI or Invoke, or call our REST endpoint that mirrors the same parameters. Teams even wrap Z Image Turbo inside serverless workers to automate daily prompt batches.