At its Dev Day, OpenAI announced updates to Codex, its software engineering agent. The main change is reusable cloud development environments, which replace isolated remote sandboxes with persistent, configurable workspaces. Developers can access these environments from a computer, phone, or the cloud, and teams can use approved settings and permissions to speed up task startup.
The update also expands Codex's role in review and security. A new code review experience in the ChatGPT desktop app lets users read summaries, explore changes, and ask about issues before sharing feedback on GitHub or GitLab. Automatic reviews can run while the user is away. Separately, Codex Security Cloud can scan entire GitHub repositories on demand or on a schedule, investigate findings, remove duplicates, and prepare fixes in the cloud, even when the developer's laptop is closed.
OpenAI also refreshed the Codex CLI with voice control and a /agents view for delegating and tracking multiple tasks. On the API side, the company introduced a Decisions API powered by Luna for real-time decision-making, and updated the Agents API to support computer use and Amazon Bedrock Managed Agents, allowing OpenAI agents to run entirely on AWS. Together, these changes position Codex less as a local assistant and more as an enterprise infrastructure component.