Zhipu AI (Z.ai) released its open-weight GLM-5.2, and some researchers claim the model matches Anthropic's Mythos in bug-finding and cybersecurity scenarios. While GLM lags behind models from Anthropic and OpenAI on more general tasks, the gap appears to have closed sharply in vulnerability-identification capabilities.
The advancement comes amid US efforts to restrict China's access to powerful models like Anthropic's Mythos and Fable, as well as the hardware needed to train and run them. The Trump administration views advanced AI capable of identifying vulnerabilities as serious national security threats, and OpenAI has similarly limited access to its recently unveiled GPT-5.6 over misuse concerns.
Because GLM-5.2 is open-weight, it can be downloaded and run by anyone on readily available hardware. That flexibility also makes it accessible to bad actors who can operate it with little oversight.