Zhipu AI Tightens Access to Coding Service Amid Computing Strain

Popularity Comes at a Price

The success of Zhipu AI's upgraded GLM-4.7 model has created an unexpected challenge - overwhelming demand that's pushing computing resources to their limits. Weekday afternoons between 3-6 PM have become particularly problematic, with users reporting slower responses and frustrating error messages.

"We're seeing server congestion that resembles rush hour traffic," explains a company spokesperson. "While we're thrilled with the adoption, we need to protect the experience for our current subscribers."

Drastic Measures Implemented

Starting January 23 at 10 AM Beijing time, Zhipu will implement strict rationing:

• Daily new sign-ups reduced by 80%
• Quotas refreshed each morning at 10 AM
• Automatic renewal users unaffected The company describes this as a temporary "supply protection" strategy while they work on expanding computing capacity.

Quality Over Quantity Approach

The decision reflects tech firms' growing struggle with computing resource allocation in the AI boom. By prioritizing existing users over rapid expansion, Zhipu hopes to:

• Maintain reliable service during peak hours
• Ensure smooth operation for programming tasks
• Buy time for infrastructure upgrades The company hasn't specified when full availability might return, promising updates based on expansion progress.

Key Points:

• GLM Coding Plan enters limited sales phase January 23
• Only 20% of current daily quota will be available
• Peak hours (3-6 PM weekdays) showing severe strain
• Existing auto-renew subscribers maintain full access
• Company prioritizing service quality over growth