Zhipu launches flagship model GLM-5.3 as China seeks Mythos-level edge in cyber defence

Chinese artificial intelligence firm Zhipu has unveiled its flagship GLM-5.3 model, claiming it outperformed Anthropic’s Mythos 5 in a key cybersecurity test. This comes as China seeks to close the gap with Western advancements in AI defence. GLM-5.3 achieved an impressive 84.5 percent success rate on CyberGym, a benchmark measuring the model's ability to identify and validate security flaws in source code.

This was higher than Mythos at 83.8 percent and OpenAI's GPT-5.6 Sol at 83.6 percent. However, GLM-5.3 did not match those foreign systems in ExploitBench, which measures the extent to which AI models can exploit vulnerabilities, scoring 54.4 percent compared to Mythos' 78 percent and GPT-5.6 Sol's 76.5 percent. Zhipu stated that GLM-5.3 had been tested with Chinese security teams against real-world codebases, uncovering 2,436 vulnerabilities in 269 projects, 1,097 of which were rated medium to high severity.

The company plans to release GLM-5.3's weights in about two weeks, after an internal safety review and hardening process. Zhipu described the model's cyber-defensive capabilities as "public goods" that developers worldwide could access and improve, emphasizing the importance of global collaboration in AI development.