On August 18, 2026, Zhipu AI released GLM-5.3, focused on improving coding, complex reasoning, and security auditing. Because the model's vulnerability-discovery ability exceeded expectations, Zhipu AI delayed the open weights and paired the release with a free security audit program.

Users on the GLM Coding Plan gradually get GLM-5.3 within their plans; model access and credit multipliers differ by tier.

New model releases often come with Coding Plan campaigns and reset card grants; the all-user reset on August 11 is one example.