OpenAI has released GPT-5.2-Codex on December 18, 2025, a variant of GPT-5.2 specifically tuned for agentic coding in Codex, including the Codex CLI (npm i -g @openai/codex). This release targets professional software engineering and defensive cybersecurity, utilizing context compaction to maintain coherence over long-running tasks. It also offers enhanced handling of large repository changes, such as refactors and migrations, and improved performance on native Windows setups. The update is now live across Codex surfaces for paid ChatGPT users, with API access planned in the coming weeks.
GPT-5.2-Codex builds on GPT-5.2’s strengths in professional knowledge work and GPT-5.1-Codex-Max’s frontier agentic coding and terminal-using capabilities, and is state-of-the-art on SWE-Bench Pro and Terminal-Bench 2.0. pic.twitter.com/weY4WHyzPf
— OpenAI Developers (@OpenAIDevs) December 18, 2025
OpenAI claims that GPT-5.2-Codex excels in long-context understanding, reliable tool calling, and native compaction while remaining token-efficient. It achieves state-of-the-art results on SWE-Bench Pro and Terminal-Bench 2.0. Additionally, its vision support enables it to interpret screenshots, diagrams, charts, and UI mocks, translating design comps into functional prototypes.
Regarding security, OpenAI highlights the React Server Components disclosures from December 11, 2025. This includes work by Privy principal security engineer Andrew MacPherson, who used Codex CLI workflows like environment setup and fuzzing to identify additional issues for responsible disclosure. OpenAI is also piloting an invite-only trusted-access track for vetted defenders, supported by safeguards such as agent sandboxing and configurable network access. Early developer discussions on Hacker News emphasize Codex’s methodical bug-finding capabilities and urge OpenAI to maintain high reasoning quality.
