Advanced capabilities require strong safeguards. That’s why access is limited to

OpenAI's Daybreak expansion separates two changes that affect security work differently. Blue gives approved defenders access to general-purpose models with adjusted system safeguards. Red provides cybersecurity-trained models, including GPT-5.6-Cyber, designed to handle specialized research and refuse fewer authorized, higher-risk requests. August 10 announcement

That distinction explains the reported 95% advanced-cybersecurity completion rate: in OpenAI's internal evaluation, it measures whether the model responds to the requests, rather than whether 95% of attempted attacks succeed. Capability evaluations were separate, and the specialized model did not outperform the general-purpose model on every test. Access combines expanded capabilities with identity verification, account-security requirements and approved-use restrictions. Evaluation and access details

“Daybreak Blue provides access to frontier general-purpose models, including GPT‑5.6 Sol, with safeguards tailored to authorized defensive security work. It is the recommended starting point for most defenders, supporting vulnerability discovery, secure code review, malware analysis, incident response, and patch validation.”

OpenAI

“GPT‑5.6‑Cyber completes 95.0% of these requests, compared with just 1.5% for GPT‑5.6 Sol, and 2.0% when used with Daybreak Blue access. It also completes more requests than GPT‑5.5‑Cyber, which completes only 57.3% of requests, addressing feedback from security researchers who encountered persistent refusals with the earlier model.”

OpenAI

“For example, we used GPT‑5.6‑Cyber to investigate V8, the JavaScript engine used by Chrome. We uncovered two previously unknown vulnerabilities that could be chained to corrupt memory and escape the V8 heap sandbox. Our researchers validated the findings and reported them to Google through coordinated vulnerability disclosure.”

OpenAI

“Under our Preparedness Framework, the GPT‑5.6 Sol model was assessed as High for cybersecurity capability and below the Critical threshold. Before launching GPT‑5.6‑Cyber, we also evaluated its frontier cyber capabilities and determined that it similarly reaches the High threshold but not the Critical threshold.”

OpenAI

“Models running with reduced safeguards carry risks beyond standard model usage, whether from misuse or misalignment. Despite these risks, we believe that democratizing access to frontier intelligence for defenders is crucial to accelerating and automating cyber defense.”

OpenAI