Data compiled by Woofun AI shows that J.P. Morgan has ranked Zhiku's GLM-5.2 as the third-largest model globally in its June 17 technology industry report. The model, which utilizes a 744 billion parameter MoE architecture with a 1 million-token context window, surpassed competitors in programming and long-term AI workflows, securing the top spot in China and third place worldwide in the Model Intelligence Index.

Despite this technical milestone, J.P. Morgan assesses that recent U.S. restrictions on Anthropic's Claude models are unlikely to drive substantial revenue growth for Zhiku. The report highlights that developers typically mix models for various tasks rather than migrating entirely, and many may utilize Chinese open-source models through local deployment or third-party platforms like OpenRouter without directly benefiting the original developers.

Furthermore, Zhiku faces significant headwinds, including a shortage of high-end inference compute power, intense price competition in the Chinese market, and long-term risks posed by U.S. export controls on distillation technology.