Z.AI Confirms Ox Alpha is a GLM Model and Plans Weight Release
Z.AI has officially confirmed that the previously mysterious "Ox Alpha" model is based on its GLM (General Language Model) architecture. Additionally, the company announced plans to release the model's weights to the public. The release of Ox Alpha's weights will allow the open-source AI community to run, fine-tune, and analyze a powerful model that previously operated under a stealth label. This move strengthens the open-source ecosystem by providing access to a high-performing model with advanced reasoning capabilities. Ox Alpha was initially released anonymously as a free frontier model featuring a 1-million-token context window and support for text, image, and video inputs. While Z.AI plans to release the weights, the exact timeline and licensing terms for the release have not yet been fully detailed.
## BACKGROUND
The GLM (General Language Model) architecture, developed by Z.AI, is a pretraining framework designed for autoregressive blank infilling. Ox Alpha had gained attention in the AI community as a high-performing, free-to-use "stealth" model on API platforms without a clear creator.