Zhipu GLM-5.3 AI: Tops Mythos 5 in Cybersecurity Test

Zhipu's GLM-5.3 AI model has outperformed Anthropic's Mythos 5 in a key cybersecurity test. The Chinese AI firm's new model scored 84.5% on CyberGym, a benchmark for identifying security flaws. This is higher than Mythos 5 at 83.8% and OpenAI's GPT-5.6 Sol at 83.6%. What's interesting is that GLM-5.3 was tested against real-world codebases, revealing 2,436 vulnerabilities across 269 projects. Of these, 1,097 were rated medium to high severity. However, the model scored only 54.4% on ExploitBench, which measures how far AI models climb the exploitation ladder. Zhipu plans to release the GLM-5.3 model publicly in about two weeks. They will allow verified users access to its most sensitive cybersecurity functions through a "trusted access" program. This aims to make advanced cyber-defense tools available to developers and smaller security teams. This development could significantly impact how software vulnerabilities are detected and addressed.