GLM-Image

GLM-Image is developed by Zhipu AI (Z.ai) and represents a breakthrough in AI image generation with its 16 billion parameter hybrid architecture combining 9B autoregressive and 7B diffusion decoder components. This design makes it the first open-source industrial-grade discrete autoregressive image generation model. The model excels particularly at rendering accurate text within images, achieving industry-leading accuracy of 0.9116 on the CVTG-2k benchmark for English text rendering. This makes it ideal for creating graphics, posters, and marketing materials that require readable embedded text. GLM-Image is available both as an open-source model on GitHub and HuggingFace, and through a commercial API with pay-per-image pricing. It supports resolutions from 512px to 2048px and includes capabilities for image editing, style transfer, and identity-preserving generation.