GLM 5.2 Beats Claude on IDOR Detection at $0.17 Per Vulnerability

GLM 5.2 from Zhipu AI scored 39% F1 on Semgrep's IDOR benchmark versus Claude Code's 32%, at approximately $0.17 per vulnerability found. Semgrep's own multimodal pipeline with purpose-built harness scaffolding still leads at 53–61% F1, suggesting model choice matters less than the surrounding tooling architecture for security tasks.