On January 8th, coinciding with Zhipu's public launch, Chief Scientist Tang Jie issued an internal memo to announce the rollout of a new-generation model, GLM-5. Rather than directly engaging with the ongoing controversies surrounding the business models of large-scale model companies or outlining specific commercialization goals for 2026, he underscored the importance of technologies that are genuinely "utilized by the public" and can assist a broader spectrum of individuals. Tang Jie highlighted that such technologies constitute significant milestones for Zhipu on its path toward achieving Artificial General Intelligence (AGI). The memo further detailed that Zhipu has successfully met its strategic objectives established at the outset of 2025, which encompassed the launch of models designated as "stabilizing the ground," "getting on the table," and "Top 1." This strategic pivot toward a comprehensive recommitment to fundamental model research is a direct response to the influence exerted by DeepSeek. On December 23rd, Zhipu unveiled its foundational model, GLM-4.7, which was subsequently open-sourced and achieved the top ranking among domestic models. In addition to the introduction of GLM-5, Zhipu is set to concentrate on pioneering entirely novel model architecture designs, more universally applicable reinforcement learning paradigms, as well as the ongoing learning and autonomous evolution of models throughout 2026.