OpenAI Launches GPT-6 Cyber for Advanced Cybersecurity Applications

OpenAI has announced the release of GPT-6 Cyber, a specialized artificial intelligence model designed to identify exploitable vulnerabilities and propose effective fixes while adhering to strict scope limitations. The model, introduced on November 5, 2026, is initially available through API and Codex access to selected Daybreak Red customers, emphasizing controlled deployment to ensure responsible use. GPT-6 Cyber delivers notable improvements in cybersecurity tasks compared to its predecessors. On ExploitGym, a platform for testing exploitation capabilities, GPT-6 Cyber achieved a 48.5% success rate, a 16.4-percentage-point increase over GPT-5.6 Cyber’s 32.1%.

Source: Blockchain.News