OpenAI CEO Sam Altman speaks at the BlackRock Infrastructure Summit in Washington on March 11, 2026.
Anna Moneymaker | Getty Images
OpenAI on Thursday announced that GPT-5.5-Cyber, a variation of its latest artificial intelligence model, will be rolled out in limited preview functionality to vetted cybersecurity teams, a month after rival Anthropic captivated investors and government officials with its Claude Mythos Preview.
The preview of GPT-5.5-Cyber is not intended to be a significant enhancement in terms of cyber capabilities, but is instead trained to be more forgiving for security-related tasks, OpenAI said in a blog post. OpenAI announced GPT-5.5 late last month.
The cyber-specific version will enable vetted teams to easily use OpenAI’s latest models for workflows such as vulnerability identification and triage, patch validation, and malware analysis, the company said. The safety features built into the commonly available GPT-5.5 model would have made it even more difficult.
“GPT‑5.5‑Cyber enables a small number of partners to explore advanced workflows where specialized access behaviors are important,” OpenAI said in a blog post.
When rolling out Mythos last month, Anthropic decided to restrict access to a select group of companies as part of a new cybersecurity initiative called Project Glasswing. Anthropic CEO Dario Amodei met with Trump administration officials to talk about the model and its potential, even after the company was placed on a Pentagon blacklist just a few weeks ago.
Federal Reserve Chairman Jerome Powell and Treasury Secretary Scott Bessent met with CEOs of major U.S. banks last month to discuss Mythos, and Vice President J.D. Vance and Bessent held a call with CEOs of major technology companies ahead of the model’s release.
WATCH: Federal agencies seek access to Mythos, prompting White House meeting with Anthropic CEO
