Z-Image Turbo TL5
Z-Image-Turbo is Tongyi-MAI's (Alibaba Group) open-weight text-to-image model, designed to generate photorealistic visuals with low computational overhead. Its 6B-parameter Scalable Single-Stream Diffusion Transformer (S3-DiT) processes text and image representations in a unified architecture, using a Qwen3-4B text encoder for prompt understanding. Distillation reduces generation to eight model evaluations, enabling fast image creation on suitable hardware; the model can run with less than 16 GB of GPU memory. It follows detailed English and Chinese prompts and can render text in both languages within generated images. Applications include product concepts, advertising visuals, posters, social graphics, and rapid creative prototyping. It generates images from text rather than editing existing images as its primary task. The weights are released under Apache 2.0 for commercial use.
Z-Image Turbo TL5
Z-Image-Turbo is Tongyi-MAI's (Alibaba Group) open-weight text-to-image model, designed to generate photorealistic visuals with low computational overhead. Its 6B-parameter Scalable Single-Stream Diffusion Transformer (S3-DiT) processes text and image representations in a unified architecture, using a Qwen3-4B text encoder for prompt understanding. Distillation reduces generation to eight model evaluations, enabling fast image creation on suitable hardware; the model can run with less than 16 GB of GPU memory. It follows detailed English and Chinese prompts and can render text in both languages within generated images. Applications include product concepts, advertising visuals, posters, social graphics, and rapid creative prototyping. It generates images from text rather than editing existing images as its primary task. The weights are released under Apache 2.0 for commercial use.