OpenAI has unveiled GPT-5.4-Cyber, a new AI model specifically designed for defensive cybersecurity. The release follows a similar announcement from Anthropic regarding its frontier AI model, Mythos.
GPT-5.4-Cyber is a fine-tuned variant of OpenAI’s GPT-5.4 model, engineered to identify security vulnerabilities in software. It has been purpose-built to lower refusal boundaries for legitimate cybersecurity tasks and includes capabilities for binary reverse engineering.
The company is also expanding its Trusted Access for Cyber program, granting access to the new model to thousands of verified security professionals. This move comes after Anthropic limited access to its Mythos model to just eleven organizations.
OpenAI’s limited release of GPT-5.4-Cyber aims to bolster defensive cybersecurity efforts with advanced AI technology.