OpenAI and CodeRabbit use GPT-6.1 Sol to strengthen code security review
OpenAI published a collaboration post with the code review tool CodeRabbit on Oct 26, in which GPT-6.1 Sol serves as the primary model for CodeRabbit's security agent, connecting behavior across files and components to surface potential security issues. The security review flow covers analysis of changes during pull request review, a Deep Scan spanning the whole repository including dependencies and component interactions, and CodeRabbit Secure, which extends investigation to linked repositories and shared components such as a shared authentication library. For findings a user selects, the flow attempts exploitability verification in an isolated sandbox without testing production systems and can draft a pull request with a proposed fix for supported issues, with people reviewing and approving, and OpenAI stressing that human reviewers remain responsible for deciding what ships. CodeRabbit's security review runs independently of its code-writing agent, the system uses other frontier models and security tooling alongside GPT-6.1 Sol, and it is positioned to complement established scanners rather than replace them, feeding findings into the pull request workflow developers already use. CodeRabbit reports that issues flagged with GPT-6.1 Sol span thousands of repositories and organizations, but the announcement discloses no vulnerability names, CVE identifiers, counts of findings or fixes, precision, recall or other quantitative benchmarks, so the phrasing cannot be read as thousands of vulnerabilities found.