The most advanced agentic coding model for professional software engineering and defensive cybersecurity.
Get started:
npm i -g @openai/codex
OpenAI is introducing GPT-5.2-Codex, the most advanced agentic coding model yet, optimized for professional software engineering and defensive cybersecurity workflows. GPT-5.2-Codex builds on GPT-5.2’s general capabilities and extends them with deeper context handling, stronger agentic coding performance, and enhanced security awareness.
What’s new:
- Professional agentic coding: optimized for long-horizon tasks like large refactors, migrations, and multi-file changes.
- Context compaction: better handling of large codebases and extended developer workflows.
- Cybersecurity focus: stronger performance on security-related tasks while being deployed with careful dual-use safeguards.
- Vision enhancements: improved interpretation of screenshots, diagrams, and UI surfaces during coding sessions.
Benefits for developers:
- Built for real-world software engineering and long-running agentic workflows.
- More dependable for complex sequences like migrations, testing pipelines, and integration tasks.
- Designed to boost productivity across IDEs, terminals, and automated CI/CD setups.
- Stronger cybersecurity capabilities help with proactive defense and secure development.
Performance:
GPT-5.2-Codex achieves state-of-the-art results on benchmarks like SWE-Bench Pro and Terminal-Bench 2.0, and markedly improves agentic coding reliability across diverse real-world software tasks.
Availability:
GPT-5.2-Codex is available today across all Codex surfaces for paid ChatGPT users. OpenAI is rolling out API access in the coming weeks and piloting invite-only trusted access for vetted professionals and organizations focused on defensive cybersecurity.
Get started:
Install the Codex CLI and select gpt-5.2-codex to begin integrating advanced agentic coding into your workflows. Share your feedback, benchmarks, and builds below!
