OpenAI rolls out advanced AI cyber model to challenge Anthropic’s Mythos

OpenAI has introduced GPT-5.5-Cyber, a model designed to find and patch cybersecurity vulnerabilities.

The model leverages OpenAI's GPT-5.5 architecture, fine-tuned for cybersecurity tasks. It can scan codebases and networks for weaknesses, suggest patches, and explain vulnerabilities. OpenAI stated that access will be restricted to approved security researchers and organizations to prevent misuse. The company described the model as a tool for both offensive and defensive security analysis.