Chinese startup Zhipu's latest AI model, GLM-5.2, can match Anthropic's most powerful AI model, Mythos 5, in cybersecurity-related tasks, WSJ reported, citing experts. Unlike Mythos 5, GLM-5.2 is open-weight, which means it can be downloaded and run on hardware operated by anyone and users can modify it without supervision. Notably, it lags behind Anthropic and OpenAI's models in other tasks.