Svmuu News: Zhipu, a leading Chinese large-language model company, recently released its open-weight model, GLM-5.2. Some researchers claim that GLM-5.2 is on par with Anthropic’s Mythos model in certain vulnerability detection and cybersecurity scenarios. Data from cybersecurity firm Semgrep shows that in certain benchmark tests, GLM-5.2 outperformed Anthropic’s Claude Opus 4.8 model, released in May. Researchers noted that when given further instructions, both Opus 4.8 and GLM-5.2 can match Mythos in terms of vulnerability detection capabilities. As an open-weight model, GLM-5.2 is available for anyone to download and run on off-the-shelf hardware.