Open Model Tops Cyber Benchmark

San FranciscoAn open-weight Chinese model outscored Claude on vulnerability detection.

GLM 5.2 from China’s Zhipu AI scored 39% F1 against Claude’s 32% on Semgrep’s security benchmark.

Zhipu AI released the weights June 16 under an MIT license, allowing self-hosting and fine-tuning without restrictions.

Sources: Hacker News