OpenAI announced significant upgrades to its Codex software engineering agent at DevDay on September 29, 2026. The centerpiece of these updates is reusable cloud development environments that enable developers to access and run Codex tasks persistently from any device, including laptops, phones, or the cloud. This allows for faster task start times and shared workspaces with controlled settings and permissions, moving away from treating cloud tasks as isolated sandboxes.

In addition to the cloud environments, the Codex command-line interface (CLI) has been refreshed to include voice control features, enabling developers to start and manage tasks hands-free. A new /agents view facilitates delegating and tracking multiple tasks simultaneously, while enhancements to editing prompts, resuming sessions, and using worktrees aim to streamline daily workflows. The terminal interface itself has been modernized for better readability during longer sessions.

OpenAI also integrated Codex into a new code review feature within the ChatGPT desktop app. This allows users to generate code summaries, examine changes, and ask Codex questions about potential problems before submitting feedback on GitHub and GitLab pull or merge requests. Automatic code reviews can perform initial assessments independently, offering continuous insights even when developers are away from their computers.

On the security front, OpenAI introduced Codex Security Cloud, a toolset for improving infrastructure security by scanning GitHub repositories on demand or scheduled intervals. It detects issues, consolidates findings, and prepares fixes in the cloud, even if users are offline. This security suite includes access to models from OpenAI’s Daybreak Blue cybersecurity initiative without requiring separate applications. Additionally, OpenAI unveiled several API enhancements, such as a real-time Decisions API powered by Luna and expanded support for Amazon’s Bedrock Managed Agents to run OpenAI agents fully on AWS.