GLM-Image

A 9B-parameter autoregressive transformer paired with a 7B-parameter diffusion decoder sits at the core of GLM-Image, an open-source AI image generator from Z.

AI (Zhipu AI).

The model is released under the Apache-2.0 license, so you can run and modify it under permissive terms.

GLM-Image is built for text to image work and image to image editing, and it reaches resolutions up to 2048x2048.

Its main selling point is text rendering: it scores 0.9116 on the CVTG-2K accuracy benchmark, which matters if your images need legible words, such as posters, signs or labels.

It also targets knowledge-intensive content, meaning scenes that depend on getting factual details right.

The glm-image.top site is an independent service that deploys and optimizes these open-source models, so it is separate from Z.

As an AI realistic image generator and art tool, GLM-Image suits designers and developers who want an open model rather than a closed one.