[OpenAI Launches GPT-6 Cyber Cybersecurity Model, Surpassing GPT-6 Astra in Vulnerability Detection Capabilities] According to a report by The Verge on November 6th, OpenAI launched GPT-6 Cyber, which scored 48.5% in the ExploitGym test, surpassing GPT-6 Astra's 42.4% and GPT-5.6 Cyber's 32.1%, and helped discover 81 previously unknown vulnerabilities in the Firefox browser. The out-of-bounds execution rate in this model test has been reduced to less than 1%, lower than the previous generation's approximately 5%, while strengthening abuse protection and intercepting high-risk attacks against critical infrastructure such as power plants and water supply facilities. GPT-6 Cyber is classified as "critical," the highest level of cybersecurity, and is currently available to some Daybreak Red customers, with plans to expand to all customers in the coming weeks.