Z Image Turbo T2I

Z Image Turbo is an open source text to image model from Alibaba Tongyi MAI, released on Hugging Face as a compact 6 billion parameter diffusion model optimized for fast inference. It generates images from natural language prompts in a handful of steps, with guidance scale tuned close to zero for the turbo variant, which makes it well suited for interactive and high throughput workflows. The model supports bilingual English and Chinese text rendering and outputs up to roughly 4 megapixels through commercial APIs, with 1024 by 1024 as a common default when run locally. Z Image Turbo is a practical choice when teams need a lightweight open model that can run on modest hardware without giving up speed or prompt adherence.

November 29, 2025

Alibaba

Text to Image (T2I)