BEIJING, June 29 — The number that matters: 39%. That is the F1 score Zhipu AI’s new GLM-5.2 open-weight model achieved on IDOR vulnerability detection, according to benchmarks from Semgrep. The figure beats Claude Code’s 32% and even outperformed Anthropic’s Claude Opus 4.8 in a bare prompt scenario.
The bottom line: a Chinese open-weight model is now trading blows with the frontier of Western cybersecurity AI. Zhipu AI, operating internationally as Z.ai, released GLM-5.2 on June 13, 2026, with open weights and API access following later the same week.
The timing is everything. The debut came just days after reports indicated Anthropic had suspended global access to its frontier models — including Mythos 5 and Fable 5 — following a directive from the U.S. Commerce Department. That move by Washington left a void.
Z.ai stepped into it. Security researchers cited by The Wall Street Journal and other outlets say GLM-5.2 demonstrates performance on par with Anthropic’s Claude Mythos in identifying complex software vulnerabilities and bugs.
For developers and enterprise security teams who lost access to Anthropic’s latest models, this is a concrete alternative — and one they can download and run themselves. GLM-5.2 may still lag behind models from Anthropic and OpenAI in certain broader, general-purpose tasks. Investors will note that distinction.
But in the narrow, high-stakes arena of cybersecurity, the gap is closing fast. Louie.ai researchers found GLM-5.2 tied with proprietary models from Anthropic in specific testing scenarios. The figure that matters for security teams is the 39% IDOR score — a benchmark that puts an open-weight model from China ahead of some of the best closed-source competitors.
By the numbers, this is not a general intelligence race. It is a surgical strike.
Z.ai targeted the cybersecurity vertical and delivered a model that security researchers are calling a game-changer for vulnerability detection. The open-weight nature of GLM-5.2 means organizations can audit the model, fine-tune it, and deploy it without sending data to a third-party API — a critical advantage for sensitive security work. The broader context is the intensifying technological competition between East and West.
With Anthropic’s frontier models effectively pulled from global markets by U.S. government directive, Z.ai’s release fills a real and immediate need. Developers who had been building on Claude Mythos now have an alternative that, at least in cybersecurity tasks, appears to match what they lost.
What to watch next: whether GLM-5.2’s cybersecurity performance translates into broader adoption, and how Western regulators respond to an open-weight Chinese model filling the gap left by their own export controls. For now, the data is on the table. The 39% F1 score speaks for itself.
{“@context”: “https://schema.org”, “@type”: “NewsArticle”, “headline”: “Zhipu AI Launches GLM-5.2 Open Model Rivaling Claude in Cyber Ops”, “description”: “BEIJING, June 29 — The number that matters: 39%. That is the F1 score Zhipu AI’s new GLM-5.2 open-weight model achieved on IDOR vulnerability detection,…”, “datePublished”: “2026-06-29 03:32:37T09:30:00+00:00”, “author”: {“@type”: “Person”, “name”: “Ava Anderson”}, “publisher”: {“@type”: “Organization”, “name”: “InfoPulse Today”, “url”: “https://newsfeed.infopulsetoday.com/”}}
