OpenAI has launched GPT-5.5-Cyber, a specialized large language model designed to help cybersecurity researchers identify vulnerabilities and automate red teaming exercises through simulated cyberattacks on critical infrastructure.

Key Points

• GPT-5.5-Cyber is available through the Trusted Access for Cyber (TAC) program, which provides expanded research capabilities to select cybersecurity experts.
• The model can generate vulnerability exploitation plans and validate them by executing simulated attacks, unlike the standard GPT-5.5 which focuses on remediation.
• It achieved an 81.9% score on the CyberGym benchmark, which evaluates LLM performance against over 1,500 historical vulnerabilities from open-source projects.
• OpenAI has implemented enhanced misuse monitoring and strict verification protocols to ensure only authorized defenders can access the model's advanced features.
• The release positions OpenAI to compete directly with Anthropic’s Claude Mythos Preview, which also offers specialized tools for software vulnerability research.

Why it Matters

This release represents a significant shift in how AI can be used to proactively secure critical infrastructure by automating complex red teaming processes. By providing researchers with validated attack simulations, OpenAI aims to accelerate the discovery of security flaws while maintaining strict safety guardrails to prevent malicious exploitation.