OpenAI Releases GPT-5.5-Cyber in Limited Preview to Security Defenders
2 min. read 
Published on
As promised by Sam Altman late last month, OpenAI today launched GPT-5.5-Cyber, its new cybersecurity-focused model built around the latest GPT-5.5 model. Notably, it’s successor of the GPT-5.4-Cyber released last month. The new model is rolling out in limited preview to defenders responsible for securing critical infrastructure to support specialized cybersecurity workflows that help protect the broader ecosystem.
Trusted Access defines model access levels
The core change is not just the model itself, but the access structure around it. OpenAI is expanding “Trusted Access for Cyber (TAC),” a framework that determines what capabilities users receive based on identity verification and approval status. It separates general users, verified defenders, and specialized security teams into different tiers.
GPT-5.5 remains the default model for general use, while GPT-5.5 with TAC is positioned for most defensive cybersecurity work. GPT-5.5-Cyber sits above both as a more permissive preview model for specialized and authorized workflows.
| Access | What changes | Intended use cases |
| GPT-5.5 (default) | Standard safeguards for general-purpose use | General-purpose, developer, and knowledge work |
| GPT-5.5 with TAC | More precise safeguards for verified defensive work in authorized environments | Most defensive security workflows, including secure code review, vulnerability triage, malware analysis, detection engineering, and patch validation |
| GPT-5.5-Cyber | Most permissive behavior for specialized authorized workflows, paired with stronger verification and account-level controls | Preview access for specialized workflows, including authorized red teaming, penetration testing, and controlled validation |
Security ecosystem integration expands
Notably, the system is also linked to wider cybersecurity ecosystem involving partners across detection, response, and network security. OpenAI says the goal is to accelerate the security cycle, from identifying vulnerabilities to validating fixes and deploying protections faster.
The announcement also includes Codex Security tools for open-source maintainers, designed to help detect and remediate vulnerabilities earlier in development. Additional safeguards are being introduced as well, including stronger authentication requirements for approved users working with sensitive capabilities.
That being said, GPT-5.5-Cyber remains in early limited preview and is not intended for broad use at this stage. Most users are expected to rely on GPT-5.5 with Trusted Access for standard defensive tasks.
Read our disclosure page to find out how can you help Windows Report sustain the editorial team. Read more
Improve this guide
User forum
0 messages