Z.ai GLM-5.3 Scores 84.5% on CyberGym Benchmark
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Z.ai said its open-source GLM-5.3 model scored 84.5% on the CyberGym benchmark, compared with 83.8% for Anthropic's restricted Mythos 5, in identifying software vulnerabilities
- Z.ai will restrict its most sensitive cybersecurity functions to verified users, creating a two-tier access model for an otherwise open-source release
Why it matters: A freely available Chinese open-source model scored 0.7 points above a gated US competitor on a cybersecurity benchmark — a notable result given Mythos 5's restricted status — while Z.ai's verified-user gate on sensitive tools reflects dual-use risk management becoming standard for frontier security AI.
Ask SkimNews
