Per Woofun AI, J.P. Morgan disclosed in a research report that Zhipu is expected to deploy its next-generation flagship model, GLM-5.5, in August. The institution anticipates the new model will feature a parameter count exceeding one trillion, marking a significant scale-up from previous iterations.

This forecast follows Zhipu's mid-June decision to open-source the GLM-5.2 model, which boasts 744 billion parameters and demonstrated robust performance in long-range programming benchmarks. The imminent release of a trillion-parameter architecture serves as a critical metric for evaluating Zhipu's ongoing technological trajectory and competitive positioning in the large language model sector.