Last updated 07-15-2026
Category:
Reviews:
Join thousands of AI enthusiasts in the World of AI!
Z-Image
Z-Image is an open-source AI image model you can run through Fooocus.one for text-to-image generation and native image editing. It is built for designers, marketers, and creators who need photorealistic visuals with accurate Chinese and English text in the same frame. Generations typically complete in eight steps, with roughly two to five seconds of wait time on consumer GPUs.
The model uses a Scalable Single-Stream DiT (S3-DiT) architecture. Text embeddings, visual semantic tokens, and image VAE tokens are concatenated into one sequence before hitting the Transformer backbone, which keeps the six-billion-parameter footprint efficient compared with dual-stream designs. A built-in Prompt Enhancer applies structured reasoning for complex instructions, from bilingual poster layouts to edits where the user's intent is only partly spelled out.
Poster designers, e-commerce teams, and bilingual publishers get a practical edge when typography has to stay sharp at small sizes. Z-Image-Edit handles bilingual editing instructions in the same interface, so you can iterate on a concept without exporting to another tool. Human preference benchmarks on Alibaba AI Arena rank it competitively against leading closed models while leading the open-source field.
Renders Chinese and English text cleanly, even at small font sizes
Eight-step generation with roughly two to five second output on consumer GPUs
Z-Image-Edit applies bilingual natural language edits without external tools
Prompt Enhancer reasons through complex instructions like puzzles or poetry
S3-DiT packs text and image tokens into one six-billion-parameter stream
Photorealistic lighting, textures, and composition tuned for poster design
Guest access on Fooocus.one includes 30 points without creating an account.
Eight-step generation delivers competitive quality against leading models cited on the site.
Z-Image-Edit handles bilingual edits in the same interface as text-to-image generation.
Strong small-font typography for posters, magazine covers, and marketing layouts.
Z-Image runs on Fooocus.one rather than a dedicated standalone product site.
Each guest generation costs five points, limiting free output to a handful of images.
Commercial license on the Basic plan requires annual billing per the pricing page.
Generation history sync requires signing in to a Fooocus.one account.
Is Z-Image free to use on Fooocus.one?
Yes. Z-Image on Fooocus.one offers guest access with 30 points and no sign-in required. Each generation costs five points, so guests can create several images before needing a paid plan or additional points.
What is Z-Image's S3-DiT architecture?
Z-Image uses a Scalable Single-Stream DiT (S3-DiT) design that feeds text, visual semantic tokens, and image VAE tokens into one unified Transformer sequence. Fooocus.one states this single-stream approach is more parameter-efficient than dual-stream alternatives for its six-billion-parameter model.
How fast is Z-Image image generation?
Z-Image targets eight-step generation. Fooocus.one reports sub-second latency on enterprise H800 GPUs, about two seconds on NVIDIA A10 hardware, and roughly two to five seconds on consumer cards like RTX 3090 or 4090.
Can Z-Image render bilingual Chinese and English text?
Yes. Z-Image is designed for accurate Chinese and English text rendering while keeping faces and overall composition realistic. Fooocus.one highlights poster layouts, magazine covers, and small-font typography as strong use cases.
What is the Prompt Enhancer in Z-Image?
The Prompt Enhancer (PE) in Z-Image uses a structured reasoning chain to interpret complex or ambiguous instructions. Fooocus.one cites examples like visualizing classical Chinese poetry or solving logic puzzles from a text prompt.
Does Z-Image support image editing?
Yes. Z-Image-Edit accepts bilingual editing instructions for in-place image changes on Fooocus.one. The page lists Text to Image and Image Editor modes, with built-in editing so you do not need a separate external editor for basic revisions.
What are the Fooocus.one paid plans for Z-Image?
Fooocus.one lists Basic at $9 per month, Plus at $18 per month, and Enterprise at $36 per month, with annual billing at $108, $216, and $432 per year respectively. Paid tiers add more points for auxiliary tools, faster generation, and commercial licensing on higher tiers.