China’s Z.ai claims it can match Mythos on cybersecurity
TL;DR
Zhipu AI, now branded as Z.ai, has released GLM-5.2 as an open-weight model. According to The Verge, some researchers say GLM-5.2 can match Mythos in specific bug-finding and cybersecurity scenarios. On broader general tasks, GLM-5.2 still trails Anthropic and OpenAI models, but the gap appears narrower in security-focused work. The important detail: open weight means many actors can download the model and run it locally on available hardware.
Nauti's Take
Open-weight security AI is not a cute benchmark story, it is a toolbox power shift. Defenders get cheaper local testers.
Teams that ignore attack surfaces will soon face opponents with similar capabilities and fewer cloud fingerprints.
Briefingshow
This is not just another benchmark claim from China. If an openly available model becomes strong at vulnerability discovery, cybersecurity shifts from controlled API access toward locally usable attack and defense automation. That combination makes export controls, model access rules, and hardware restrictions much harder to enforce.
Sources