OpenAI Codex Gets Computer Use, In-App Browser, and Image Gen
OpenAI ships a sweeping Codex desktop update with background computer use, a built-in Atlas browser, image generation, 111 new plugins, and memory — framing it as the first milestone toward a developer superapp.
Image by OpenAI
OpenAI Codex Gets Computer Use, Browser, and Image Gen in Sweeping Update
OpenAI on Thursday pushed a major update to its Codex desktop app, adding background computer use, an in-app browser powered by Atlas, image generation via gpt-image-1.5, persistent memory, 111 new plugins, and long-running automation scheduling — signaling a deliberate push toward what the company calls a developer-first "superapp."
The release, announced April 16, was framed by Thibault Sottiaux, head of Codex, as the start of a public assembly of that superapp vision. "We're building the super app out in the open," Sottiaux said in a press briefing. "This release is about developers. In the future, we will broaden it up to a wider audience."
The update arrives as Codex momentum accelerates. OpenAI disclosed that Codex now has 3 million weekly active users — a 5x increase in three months — with 70% month-over-month usage growth. The company also noted that 80% of OpenAI's internal staff use Codex.
Background Computer Use
The headline feature is computer use. Codex can now see, click, and type into other macOS applications using its own cursor, while the user continues working simultaneously. Multiple agents can run in parallel without interfering with the user's own applications.
OpenAI says this is particularly useful for testing frontend changes, iterating on apps, and operating tools that don't expose an API. The implementation was built on technology from OpenAI's acquisition of Sky Applications Incorporated — the team originally behind Apple Shortcuts and Workflow — last fall.
The stated differentiator is a background execution layer that lets agents operate apps without hogging system resources, so the developer and the agent can use the machine at the same time. Computer use is macOS-only at launch. EU and UK users will receive access in a future rollout.
In-App Browser and Image Generation
Codex now ships with a built-in browser built on OpenAI's Atlas technology — the same engine behind the standalone ChatGPT Atlas browser released last fall. The browser supports direct page commenting: you annotate a specific element on a page and instruct Codex to change it. OpenAI demonstrated a developer fixing a cut-off graph axis label through a single annotation.
"This is useful for frontend and game development today, and over time we plan to expand it so Codex can fully command the browser beyond web applications on localhost," the company said.
Alongside the browser, image generation powered by gpt-image-1.5 is now native to Codex, removing the need to context-switch to the ChatGPT app. The feature targets frontend developers who need quick mockups, product concepts, or in-context design assets. Codex can also use screenshots during task execution to verify it is on the right track with a request.
111 Plugins, Memory, and Persistent Automation
OpenAI released a curated collection of 111 new Codex Plugins, combining skills, app integrations, and MCP server connections. These extend what agents can pull into context and which tools they can invoke from inside a single session.
Two memory features are now in preview. The first lets Codex recall preferences, corrections, and context accumulated from previous sessions, allowing future tasks to complete faster and to a higher quality standard without re-explaining the same project details. The second enables proactive suggestions at session start: Codex can surface open Google Docs comments awaiting attention, pull relevant context from Slack and Notion, and generate a prioritized action list.
Automations have also been extended to re-use existing conversation threads, preserving context built up across prior sessions. Codex can now schedule future work for itself and "wake up automatically to continue on a long-term task, potentially across days or weeks," per OpenAI.
Developer Workflow Additions
Beyond the headline features, the release adds support for addressing GitHub pull request review comments directly inside the app, multiple concurrent terminal tabs, and remote devbox connections over SSH — the last of which is currently in alpha. A new summary pane lets developers track agent plans, sources, and artifacts in one place. Files can now be opened in the sidebar with rich previews for PDFs, spreadsheets, slides, and documents.
The release also marks the first time Codex for Mac supports Intel hardware, significantly widening the installed base that can run the desktop agent natively.
What's Unconfirmed
OpenAI has not published benchmark numbers or independent latency data for the computer use implementation. The claim that the background execution layer avoids degrading system performance has not been independently verified under sustained multi-agent load. The memory feature remains a "preview" with no timeline for general availability, and no date has been set for EU and UK access to computer use or memory. The broader superapp — a unified ChatGPT + Codex + Atlas application confirmed last month by OpenAI — is not shipping today; the company gave no release window.
