OpenAI released GPT-5.4-Cyber to a restricted group of vetted cybersecurity professionals. The model provides specialized capabilities for cyber defense while operating under strict safety safeguards. This limited rollout prevents misuse by malicious actors. Practitioners gain a high-reasoning tool for threat detection without risking the proliferation of offensive AI capabilities.