China's Zhipu AI claims to have developed the GLM-5.2 model, matching Anthropic's Mythos in cybersecurity tasks. However, it ...