GLM Image T2I

GLM Image is Zhipu AI's open source text to image model, also branded internationally as Z.ai. It produces images from natural language prompts with an emphasis on accurate instruction understanding, structured composition, and strong text rendering inside images. The model uses a hybrid architecture that pairs a 9 billion parameter autoregressive language model for global layout and semantic control with a 7 billion parameter diffusion decoder for fine detail and text stroke quality. GLM Image is aimed at knowledge and layout heavy outputs like posters, presentation style graphics, infographics, and other designs that require readable typography and consistent placement. It generates images up to 2048 by 2048 with custom sizes in multiples of 32 and is distributed with open weights and a hosted API.

January 13, 2026

Zhipu AI

Text to Image (T2I)