OpenAI has launched GPT‑5.2‑Codex, its most advanced coding model optimized for professional software engineering and cybersecurity applications. This version improves on the previous GPT‑5.2 with enhanced capabilities for complex coding tasks, such as long-horizon code changes and better performance in Windows environments. Notably, it strengthens cybersecurity functions, making it a more effective tool for identifying and mitigating vulnerabilities. While it doesn't yet meet the 'High' cybersecurity capability level, OpenAI is preparing for future models to reach that threshold. The model excels in real-world software engineering benchmarks, achieving top scores on SWE-Bench Pro and Terminal-Bench 2.0. These tests simulate realistic coding environments where the model must generate patches and perform tasks like compiling code and setting up servers. GPT‑5.2‑Codex shows improved long-context understanding and can maintain context over extended coding sessions, allowing it to handle large refactors and migrations more reliably. Its enhanced vision capabilities also allow it to interpret complex technical diagrams and screenshots effectively. A recent example of its cybersecurity prowess came when a researcher using GPT‑5.1‑Codex‑Max identified vulnerabilities in React Server Components. This incident underscores the growing importance of AI tools in finding and addressing security flaws in software. OpenAI is implementing more safeguards with the release of GPT‑5.2‑Codex to manage the risks that come with its advanced capabilities, particularly in cybersecurity. The model is currently available to paid ChatGPT users, with plans for broader access to API users and vetted professionals focused on defensive cybersecurity.