OpenAI announces the update of the GPT-5.5-Cyber model (new), which achieves a 85.6% score on the CyberGym benchmark compared to 81.9% in its initial version. Codex also obtained a new security plugin.

→ View original post on X — @testingcatalog