OpenAI has upgraded Codex from a basic cloud-based coding tool to a full-fledged system of reusable development environments that can be accessed from various devices. OpenAI introduced new capabilities for Codex at its Dev Day on Tuesday, enabling developers to switch tasks between their laptops, mobile devices, IDEs, and the cloud.
This enhancement is part of a larger set of improvements for Codex, which includes a refreshed Codex Command Line Interface (CLI), a new code review experience, and tools focused on strengthening infrastructure. Additionally, OpenAI is rolling out several other API updates to enhance Codex’s functionality.
Now, developers can run Codex wherever they need it–on a computer, remotely from a phone, or in the cloud. The reusable environments allow teams to start tasks more quickly and provide a shared workspace with approved settings and permissions.
The Codex CLI is being updated to allow developers to start and direct tasks using their voice. The Codex CLI is being updated to allow developers to start and direct tasks using their voice. Alongside this voice command capability, there are improvements to everyday tasks such as editing prompts, resuming sessions, and using worktrees. These updates contribute to a cleaner interface that makes longer sessions easier to read.
In addition, Codex now integrates with a new code review experience within the ChatGPT desktop app. This feature enables users to read summaries, review code changes, and consult Codex about potential issues before giving feedback on GitHub pull requests or GitLab merge requests. Automatic code reviews let Codex perform an initial review while users step away from their computers.
On the security side, OpenAI is launching Codex Security Cloud, a suite of tools designed to help developers secure their infrastructure. These tools allow developers to scan entire GitHub repositories on demand or regularly. Codex will analyze findings, eliminate duplicates, and prepare fixes in the cloud, even when the user is inactive and their laptop is closed. Notably, developers can access the models offered through OpenAI’s cybersecurity initiative, Daybreak Blue, without needing to submit a separate application.
OpenAI introduced a new Decisions API for real-time decision-making that uses Luna to answer user-defined questions. The updated Agents API now supports computer usage, allowing developers to create OpenAI agents that run entirely on AWS using Amazon’s Bedrock Managed Agents.
