Zhipu launches GLM-5.3 model as China seeks Mythos-level edge in cyber defence

Zhipu launches GLM-5.3 model as China seeks Mythos-level edge in cyber defence

Zhipu's GLM-5.3 beat Anthropic's Mythos in code vulnerability detection (84.5% vs 83.8%), showing China's AI security push. But it lags badly in real-world exploitation (54.4% vs 78%), exposing the gap between finding flaws and weaponizing—key for enterprises evaluating threat modeling AI.