OpenAI has released GPT-5.5-Cyber Full Version, scoring 85.6% surpassing Mythos 5

According to the latest Beating Monitoring report, OpenAI has announced the upgrade of the Daybreak network defense toolkit and officially launched the GPT-5.5-Cyber full version model targeted at defenders. In the evaluation of the CyberGym benchmark test, assessing the intelligent agent's ability to reproduce known vulnerabilities, GPT-5.5-Cyber achieved the highest single-model score of 85.6%, surpassing GPT-5.5 at 81.8% and the competitor Mythos 5 at 83.8%.

As the widespread adoption of AI has significantly increased the efficiency of vulnerability discovery, the bottleneck on the defense side has shifted from "finding vulnerabilities" to "automatic remediation." To translate detection results into actual fixes, the accompanying Codex Security plugin has been upgraded, allowing developers to automatically analyze and generate repair patches within the Codex environment. Since its preview release in March, Codex Security has scanned over 30 million commits cumulatively, automatically confirming the remediation of 500,000 security flaws. The latest version supports integration with CodeQL queries and exporting SARIF standard files, enabling the interception of potential vulnerabilities before software deployment.

Simultaneously, OpenAI, in collaboration with organizations such as Trail of Bits and HackerOne, has launched the Patch the Planet open-source security project. This initiative provides ChatGPT Pro subscriptions and API quotas for over 30 mainstream open-source projects like cURL and Go, with a expert team manually verifying patches to ease the burden on maintainers. In the Daybreak partner program, initial participants including Palo Alto Networks and Wiz security vendors are committed to implementing advanced defense capabilities in commercial security products.