Z.ai Drops GLM-5.3 Weights With Mandatory Security Reviews for $10B Hyperscalers

Chinese AI lab Z.ai published the weights for GLM-5.3, a 753-billion-parameter mixture-of-experts model featuring a 1 million-token context window. The updated GLM-5.3 license mandates that hosting entities generating more than $10B in aggregate revenue over 12 consecutive months must pass a Z.ai security review before commercial use. The model achieved a self-reported 84.5% score on the CyberGym vulnerability discovery benchmark and reportedly found 2,436 vulnerabilities across 269 open-source projects. Deploying the model locally remains hardware-intensive, requiring 245GB of memory for 2-bit quantization and 810GB for 8-bit quantization.