OpenAI responded to Claude Mythos with a new GPT‑5.5‑Cyber that can discuss cyberattacks and exploits, although it was released with a one-month delay.

OpenAI has released a new AI model for cybersecurity

The company announced the launch of GPT‑5.5‑Cyber – a specialized model designed specifically for information security work. It is essentially an “analogue” of the popular Anthropic Mythos product, but with unique features and limited access.

Key details
Parameter Information Model GPT‑5.5‑Cyber Purpose Vulnerability detection, incident prioritization, patch verification, and malicious code analysis Release date Less than a month after the release of GPT‑5.4‑Cyber Access Only participants in the Trusted Access for Cyber (TAC) program

What’s new compared to the previous version?
Developers noted that the changes “are not considered a major update.” The model retains the same core functions but is optimized for broader cybersecurity use cases.

Trusted Access for Cyber (TAC)
Since the launch of GPT‑5.4‑Cyber, OpenAI has expanded the TAC program to:

- Several thousand vetted individual experts,
- Several hundred teams responsible for protecting critical software.

Thus, access to GPT‑5.5‑Cyber is granted only to qualified specialists and organizations that have passed rigorous vetting.

OpenAI Comment
> “The cybersecurity ecosystem is vast, and GPT‑5.5 with GPT‑5.5‑Cyber play different roles in meeting the needs of organizations and researchers. For most teams, GPT‑5.5 via TAC is the most effective and widely applicable model for legitimate cybersecurity work with reliable safeguards against misuse,” said OpenAI representatives.

In this way, the new model continues to strengthen the company’s position in specialized AI solutions for information protection, but does so through a strict access control system.