OpenAI has recently unveiled GPT 5.2, along with updates to its Codex coding model, enhancing its capabilities for a variety of agentic use cases. GPT-5.2-Codex is described as the most advanced agentic coding model by OpenAI, specifically designed for complex real-world software engineering tasks. This model has been optimized for long-horizon work with agents and boasts stronger cybersecurity capabilities.
The development of GPT-5.2-Codex marks a significant advancement in AI support for software engineering and specialized domains like cybersecurity. OpenAI emphasizes the model’s ability to assist developers and defenders in tackling intricate, long-term projects and enhancing tools for responsible security research.
Enterprises now have access to the new Codex model through various channels, including Codex surfaces for paid ChatGPT users, with plans to enable access to GPT-5.2-Codex for API users in the near future. Additionally, OpenAI is piloting a program for trusted users to access more permissive models tailored for defensive cybersecurity work, striking a balance between accessibility and security.
In terms of cybersecurity, OpenAI highlights GPT-5.2-Codex as its most robust cybersecurity model to date. The company acknowledges the need for a deployment strategy that accommodates future growth and supports defensive cybersecurity measures. Through rigorous testing on benchmarks like Capture-the-Flag (CTF) evals, CVE-Bench, and Cyber Range, GPT-5.2-Codex has demonstrated exceptional performance, showcasing its effectiveness in cybersecurity tasks.
OpenAI’s commitment to enhancing cybersecurity capabilities is evident in its cybersecurity deployment project. Users of GPT-5.1-Codex-Max identified a source code exposure vulnerability in React, underscoring the model’s potential for real-world vulnerability research. With the advancements in cybersecurity capabilities in GPT-5.2-Codex, OpenAI is focused on deploying frontier models responsibly, ensuring that defensive cybersecurity tools are in place to mitigate potential risks.
GPT-5.2-Codex is praised for its versatility in business tasks and workflows, with its enhanced capabilities for coding agents. The model excels in long-horizon work, offering strong performance in handling extensive code changes and improved performance on Windows platforms. Through benchmark testing, GPT-5.2-Codex has proven to be more accurate and reliable in completing complex coding tasks.
Since its preview release in May, Codex has revolutionized agentic and vibe coding in the enterprise AI builder space. Together with other coding agents like Windsurf, Cursor, and Claude Code, Codex has shifted the landscape of AI-powered coding tools, empowering users to undertake complex coding projects with ease and efficiency.