OpenAI Super App Takes Shape: Codex Gets Computer Use, Browser, and Image Gen

OpenAI announced today it has updated its Codex desktop app with computer use, an in-app browser, image generation, and over 90 new plugins.

Almost one year after initially launching Codex, the company now says more than 3 million developers use it every week. And now the idea, OpenAI says, is to let them use “Codex for (almost) everything.”

With background computer use, Codex can now see your screen, move its own cursor, and click and type inside any Mac application. Multiple agents can run simultaneously without interrupting whatever you’re doing on the side. OpenAI says it’s most useful for frontend iteration, app testing, and workflows that don’t expose an API.

The in-app browser lets users comment directly on pages to give the agent precise instructions. OpenAI says it’s aimed at frontend and game development today, with plans to expand full browser control over time. Image generation, powered by gpt-image-1.5, is now built into the same workflow—no API key needed, and with usage covered by a ChatGPT account.

It also features 90+ new plugins, including integrations with Atlassian Rovo, CircleCI, CodeRabbit, GitLab Issues, the Microsoft Suite, and Neon by Databricks. They combine skills, app integrations, and MCP servers to extend what Codex can access and act on across a developer’s existing toolset.

On the workflow side, the app now supports multiple terminal tabs, GitHub PR review comment handling, SSH connections to remote devboxes (in alpha), and a summary pane that tracks agent plans, sources, and artifacts. Files open directly in the sidebar with rich previews for PDFs, spreadsheets, slides, and docs.

OpenAI says the goal is to reach “a level of quality previously only possible through extensive custom instructions.”

There’s also a new proactive mode. Using context from connected plugins, memory, and active projects, Codex can suggest where to start a work day or resume a previous task—pulling open Google Docs comments, relevant Slack threads, Notion pages, and codebase context into a prioritized action list.

The feature set covers a lot of the same ground as OpenClaw, the open-source agent framework that went viral in early 2026. OpenClaw was built by Austrian developer Peter Steinberger to run persistent agents locally, connected to messaging apps, files, browsers, and shell commands. It accumulated 60,000 GitHub stars in 72 hours and drew comparisons to a personal AI operating system.

Steinberger joined OpenAI in February to lead personal agent development after Sam Altman, Mark Zuckerberg, and Satya Nadella all reached out following OpenClaw’s rise. The project moved to an open-source foundation with OpenAI as financial sponsor.

Before the OpenAI hire, Anthropic had sent Steinberger a trademark complaint over the original name “Clawdbot”—a dispute that triggered two chaotic rebrands and, according to observers, accelerated his move to OpenAI. OpenClaw had been running primarily on Anthropic’s Claude models at the time.

The more direct mainstream, closed source coding-tool rival to OpenAI’s Codex is Anthropic’s Claude Code, a terminal-based agentic coding assistant that reads entire codebases, edits files, runs tests, and commits to GitHub. Anthropic also introduced its own computer use feature for Claude in March, available as a research preview for Pro and Max subscribers on macOS.

Codex packages these capabilities differently—computer control, browsing, image generation, and coding in one desktop app, tied to a ChatGPT account. OpenAI describes the direction as an effort on “narrowing the gap between what people can imagine and what they can build.”

The update is rolling out today to Codex desktop users signed in with ChatGPT. Personalization features and computer use are not yet available in the EU or UK.