CodeRabbit uses GPT-6.1 Sol to connect code review with wider security investigation

The vendor case study describes cross-repository analysis, isolated verification and proposed fixes that remain subject to human approval.

CodeRabbit is using GPT-6.1 Sol as the primary model in its security agent, according to an October 26 OpenAI case study. The system aims to connect individual code changes with behavior elsewhere in an application, where shared components and dependencies can determine whether a change introduces a security risk.

The account describes pull-request review, broader repository investigation through Deep Scan and linked-repository investigation through CodeRabbit Secure. For findings selected by a user, supported workflows can attempt verification in an isolated sandbox and draft a repair pull request for human review. The case study says this does not involve testing production systems.

The useful distinction is between identifying suspicious code and showing why it matters in a complete system. A shared authentication function may appear sound in isolation while one caller applies its result incorrectly. Understanding that relationship requires following information and permissions across component boundaries.

Verification and repair add separate decisions. A reviewer needs to understand the conditions required for a finding, whether the proposed change addresses those conditions and whether the application still behaves as intended afterward. Producing a patch is one step in that process rather than its endpoint.

The case study presents a vendor-described workflow, without an independent accuracy comparison or a universal vulnerability-detection rate. Its practical contribution is the connection between investigation, bounded validation and reviewable remediation. For development teams, those connections can make a finding easier to act on when the evidence, proposed fix and final approval remain visible in familiar review tools.

Sources

- [GPT-6.1 Sol helps CodeRabbit strengthen security across complex codebases](https://openai.com/index/gpt-6-1-sol-coderabbit-code-security/) — OpenAI