GPT-5.5-Cyber debuts in limited preview, taking aim at Anthropic's Mythos

OpenAI introduced GPT-5.5-Cyber, a security-tuned variant of GPT-5.5 available via its Trusted Access for Cyber (TAC) program launched in February for vetted defenders of critical infrastructure. The model targets PoC writing and offensive/defensive simulations — directly challenging Anthropic's Mythos in the cybersecurity model space.