OpenAI has unveiled GPT-5.2-Codex, a cutting-edge mannequin optimized for agentic coding and enhanced cybersecurity duties. The discharge highlights breakthroughs in dealing with advanced software program engineering and vulnerability detection.
GPT-5.2-Codex tops SWE-Bench Professional with 56.4% accuracy, outperforming GPT-5.2 at 55.6% and GPT-5.1 at 50.8%. On Terminal-Bench 2.0, it scores 64.0%, surpassing prior variations like GPT-5.2’s 62.2%. These good points stem from improved long-context dealing with, software use, and native compaction for prolonged coding periods.
BenchmarkGPT-5.2-CodexGPT-5.2GPT-5.1-Codex-MaxSWE-Bench Pro56.4% 55.6%50.8% Terminal-Bench 2.064.0% 62.2% 58.1%
The mannequin excels in skilled Seize-the-Flag challenges, exhibiting sharp functionality jumps over predecessors. It helps fuzzing, check atmosphere setup, and assault floor evaluation, accelerating defensive workflows. OpenAI notes stronger safeguards regardless of dual-use dangers, staying under “Excessive” cyber functionality per its Preparedness Framework.
A researcher utilizing GPT-5.1-Codex-Max uncovered flaws in React Server Parts whereas probing CVE-2025-55182, a essential RCE (CVSS 10.0) mounted December 3.
This led to CVE-2025-55183 (supply code publicity, CVSS 5.3), CVE-2025-55184, and CVE-2025-67779 (DoS, CVSS 7.5), disclosed December 11. The method concerned iterative prompting, native setups, and fuzzing, as diagrammed in shared visuals.
GPT-5.2-Codex launches for paid ChatGPT Codex customers, with API entry quickly. An invitation-only pilot targets vetted cybersecurity execs for defensive duties like red-teaming. OpenAI pairs advances with mannequin safeguards and group collaboration to curb misuse.
This evolution equips builders and defenders in opposition to rising threats in codebases and infrastructure. React patches urge upgrades to variations like 19.0.3+.
AI-Powered ISO 27001, SOC 2, NIST, NIS 2, and GDPR Compliance Guidelines => Begin for Free
