On February 16th, reports emerged indicating a significant surge in global demand for GLM-5. Concurrent access volumes have surpassed the planned thresholds, resulting in service queuing, delayed responses, and lags, all of which have collectively degraded the user experience. Despite Zhipu's (02513.HK) repeated efforts to expand its domestic chip clusters and the limited sales of the GLM Coding Plan package, the demand remains unmet.

To tackle this challenge, Zhipu has initiated a recruitment drive for ‘Computing Power Partners.’ This initiative encompasses three key strategies:

1. Collaborating with chip manufacturers to unlock core technology interfaces and jointly enhance the foundational architecture of GLM-5;
2. Forming partnerships with computing power providers and inference service providers to construct a high-concurrency, low-latency inference network;
3. Exploring alternative avenues for computing power collaboration.