GPT-5.5-Cyber to Roll Out in Next Few Days for "Critical Cyber Defenders," Says OpenAI CEO

Earlier this month, OpenAI introduced GPT-4-Cyber, its own frontier cybersecurity model. That came shortly after the announcement of Anthropic’s Claude Mythos cybersecurity model, which is part of Project Glasswing. At the time, OpenAI confirmed that it would initially only be available to vetted security vendors, approved organizations, and selected researchers under its Trusted Access for Cyber (TAC) program.

OpenAI teases GPT-5.5-Cyber with limited details

Now that GPT-5.5 is here, OpenAI is gearing up for the launch of GPT-5.5-Cyber, an upgraded cybersecurity model. The confirmation came straight from CEO Sam Altman on X, as he said it will be available first to “critical cyber defenders in the next few days” (via The Verge)

Altman further added, “we will work with the entire ecosystem and the government to figure out trusted access for cyber; we want to rapidly help secure companies and infrastructure.”

As of now, it’s unclear who exactly will get access to the new cybersecurity model. OpenAI previously said that with its cybersecurity model, it wants to support defensive workflows like vulnerability analysis, code review, and reverse engineering of compiled software.

Notably, Anthropic’s Claude Mythos was reportedly accessed by unauthorized groups days after launch, though no harm was reported. After reports surfaced, the company said it is looking into the matter, stating, “We’re investigating a report claiming unauthorized access to Claude Mythos Preview through one of our third-party vendor environments.”

Following that report Reuters reported that the Reserve Bank of Australia is actively engaging with regulators, government bodies, and industry players as Claude Mythos gains attention. The Reserve Bank of New Zealand also reportedly stepped in to coordinate with both domestic and Australian counterparts over what it described as a “developing risk.”