OpenAI released GPT-5.5-Cyber and GPT-5.5 to a restricted group of verified security researchers. These models specifically target vulnerability research and critical infrastructure protection. Access remains limited to prevent misuse. This targeted release allows defenders to automate threat detection without exposing high-risk capabilities to the general public.