OpenAI turns Codex into an always-on coding agent that watches your screen

OpenAI is massively expanding its developer tool Codex: the AI can now control a Mac on its own, generate images, remember preferences, and keep working on tasks autonomously for weeks. The move takes direct aim at Anthropic's Claude Code.

OpenAI has released a comprehensive update for Codex that pushes the AI coding assistant well beyond its previous role as a terminal and editor tool. The biggest new feature is "background computer use:" Codex can now operate any app on the user's computer by seeing the screen, clicking, and typing with its own cursor, according to OpenAI.

Multiple agents can run in parallel on a Mac without interfering with whatever the user is doing in other apps. OpenAI says this is especially useful for developers iterating on front-end changes, testing apps, or working with programs that don't have an API. The feature is only available on macOS for now.

The Codex app also now includes a built-in browser where users can comment directly on web pages to give the agent specific instructions. Right now, this is mainly geared toward front-end and game development. OpenAI plans to expand the browser feature so Codex can fully control the browser beyond local web apps.

Codex grows into a full software development companion

The update also adds much broader support across the software development workflow. Codex can now edit GitHub review comments, run multiple terminal tabs at once, and connect to remote devboxes via SSH in alpha, for example.

OpenAI has expanded the automation side too: existing conversation threads can be reused so previously established context carries over. Codex can schedule itself for future tasks and wake up on its own to continue working on long-term projects, "potentially across days or weeks," according to OpenAI.

Teams can use these automations for everything from processing open pull requests and tracking tasks to monitoring conversations in Slack, Gmail, and Notion.

Image generation and over 90 new plugins

Codex now uses gpt-image-1.5 for image generation. Combined with screenshots and code, this lets teams create and iterate on product concepts, front-end designs, mockups, and game graphics within the same workflow.

OpenAI has also shipped more than 90 additional plugins that bundle skills, app integrations, and MCP servers. New additions include Atlassian Rovo for JIRA management, CircleCI, CodeRabbit, GitLab Issues, Microsoft Suite, Neon by Databricks, Remotion, Render, and Superpowers. The plugins give Codex more ways to pull context from different tools and act on it directly.

Rollout starts now, but some features are region-locked

The updates are rolling out immediately to Codex desktop app users logged in with a ChatGPT account. Personalization features like context-aware suggestions and memory will follow soon for Enterprise, Edu, EU, and UK users. The computer use feature is limited to macOS for now and will reach EU and UK users at a later date.