On November 6, U.S. tech media outlet The Verge reported that OpenAI has launched GPT-6 Cyber. The model scored 48.5% on the ExploitGym test, surpassing GPT-6 Astra's 42.4% and GPT-5.6 Cyber's 32.1%, and helped uncover 81 previously unknown vulnerabilities in the Firefox browser. Its out-of-bounds execution rate dropped to under 1%, down from roughly 5% in the previous generation. Meanwhile, abuse protections have been strengthened, enabling it to intercept high-risk attacks targeting critical infrastructure such as power plants and water supply facilities. GPT-6 Cyber has been classified at the highest tier in the cybersecurity field—"Critical level"—and is currently available to select Daybreak Red customers, with plans to expand access to all customers in the coming weeks.