Software engineers now have a powerful new AI assistant with the release of GPT‑5‑Codex. The tool can autonomously create and review code, detect software flaws, and integrate with multiple platforms including IDEs like VS Code, GitHub, the cloud, and ChatGPT mobile apps.

Performance improvements are a major focus of the upgrade. GPT‑5‑Codex reportedly reduces task completion times by up to 90% for certain cloud-based operations while providing accurate code comments and supporting website creation.

Safety remains a key priority. All tasks are executed in a sandboxed environment with network access disabled by default. Developers can enable web access for project iteration and autonomous code reviews, but safeguards are in place to prevent misuse.

High-risk domains receive extra protections. OpenAI flags the AI as highly capable in biological and chemical contexts, monitoring its use to prevent harmful activities and ensuring responsible deployment.

User accessibility spans multiple subscription tiers. ChatGPT Plus, Pro, Business, Edu, and Enterprise subscribers can leverage GPT‑5‑Codex for both routine and complex coding tasks.

Autonomy and convenience are balanced with caution. The system’s ability to act independently speeds development but requires careful oversight to avoid potential security or misuse concerns.

For more details on GPT‑5‑Codex, visit the full NotebookCheck report. Source