OpenAI expands Codex beyond coding with computer use, browser work, memory, and automations

OpenAI used April 16, 2026 to reposition Codex from a coding-focused assistant into a broader software-workflow agent. The update adds background computer use on macOS, an in-app browser, image generation with gpt-image-1.5, memory, context-aware suggestions, and automations that can continue work across days or weeks.

That matters because Codex is no longer framed only as a tool for writing code faster. OpenAI is now pitching it as a product that can move between coding, testing, review, coordination, and repeatable follow-up work in one workspace. For ClawNewbie readers, this is the kind of release that can change how Codex compares with editor-first tools, terminal-heavy agents, and broader coding stacks.

OpenAI used April 16, 2026 to reposition Codex from a coding-focused assistant into a broader software-workflow agent. The update adds background computer use on macOS, an in-app browser, image generation with gpt-image-1.5, memory, context-aware suggestions, and automations that can continue work across days or weeks.

That matters because Codex is no longer framed only as a tool for writing code faster. OpenAI is now pitching it as a product that can move between coding, testing, review, coordination, and repeatable follow-up work in one workspace. For ClawNewbie readers, this is the kind of release that can change how Codex compares with editor-first tools, terminal-heavy agents, and broader coding stacks.

According to OpenAI's April 16, 2026 post, Codex now supports several capabilities that extend it beyond a narrow coding workflow:

• background computer use so Codex can see, click, and type with its own cursor on your computer
• an in-app browser so users can comment directly on pages and iterate on frontend work faster
• image generation with gpt-image-1.5 for mockups, concepts, and game or app visuals
• memory and context-aware suggestions so Codex can carry forward preferences and previous corrections
• automations that can reuse existing threads and wake up later to continue long-running work
• deeper developer workflow support such as reviewing pull requests, viewing multiple files and terminals, and connecting to remote devboxes over SSH in alpha

OpenAI also said more than 90 additional plugins are rolling out, giving Codex more ways to gather context and take action across tools.

The important shift is not any single feature. It is the product direction underneath them.

Before this announcement, many readers would have evaluated Codex mainly against the question, How good is it at generating or editing code? After the April 16, 2026 update, the more useful question becomes, How much of the software workflow can Codex now operate end to end?

That changes the evaluation frame in at least three ways:

1. Codex is moving closer to browser-aware and desktop-aware workflows instead of staying inside text-only coding loops.
2. Memory and automations make Codex more relevant for ongoing projects, not just one-off prompts.
3. Review, multi-terminal support, and SSH connections make Codex easier to compare as a workflow hub rather than a single-surface assistant.

That broader framing is why this news story matters to readers already using the site's Codex tool page, Gemini CLI vs Codex comparison, Claude Code vs Codex comparison, and Cline vs Codex comparison.

Several details in OpenAI's announcement are especially relevant for evaluation instead of just product curiosity.

Computer use and browser work

OpenAI said Codex can now operate the computer alongside the user and includes an in-app browser where users can comment directly on pages. That directly affects how readers should think about frontend iteration, UI testing, and tasks that previously required an API or manual clicks.

If that rollout proves stable in practice, it narrows part of the gap between coding assistants and broader workflow agents.

Memory and longer-running automations

OpenAI said Codex can now reuse existing conversation threads, preserve context over time, wake up automatically for future work, and remember useful context through a preview of memory.

That matters because many coding-tool decisions break down after the first session. A tool that helps once but loses context later often creates more work than it saves. Memory and automation features are OpenAI's answer to that problem.

More workflow depth, not just more chat

OpenAI highlighted pull-request review support, multiple terminal tabs, direct file previews, and SSH access to remote devboxes in alpha. Those are practical workflow details, not just demo features.

For ClawNewbie readers comparing stacks, these are the signals that Codex is being positioned as a working environment with agent support, not only as a model wrapped in a coding UI.

OpenAI said on April 16, 2026 that the update is rolling out to Codex desktop app users signed in with ChatGPT. The company also said personalization features such as memory and context-aware suggestions would roll out to Enterprise, Edu, and EU and UK users later, while computer use is initially available on macOS and will also reach EU and UK users later.

That means readers should avoid assuming every feature is instantly available in every region or plan tier. The product direction is clear, but rollout details still matter.

This announcement gives the /news hub a strong entry point because it bridges directly into multiple live coding-cluster assets:

• readers new to Codex can start at Codex
• readers comparing terminal-heavy and broader workflow tools can branch into Gemini CLI vs Codex, Claude Code vs Codex, and Cline vs Codex
• readers still deciding across the broader category can move to Best AI Coding Tools 2026

The release also creates obvious follow-up angles for future pages:

• a compare refresh focused on whether browser work and computer use materially change Codex versus editor-first rivals
• a resources page on long-running agent workflows, automation setup, or memory-aware operating patterns

OpenAI's April 16, 2026 announcement is important because it changes the shape of the Codex evaluation, not just the feature list. Codex is now being presented as a system that can write code, review work, use a browser, operate parts of the computer, remember context, and keep longer-running tasks moving over time.

If you have not looked at Codex since it was primarily a coding tool, the old mental model is now incomplete.

What did OpenAI announce for Codex on April 16, 2026?

OpenAI said Codex now adds computer use, an in-app browser, image generation, memory, context-aware suggestions, automations, more plugins, and deeper support for PR review, file previews, multiple terminals, and SSH workflows.

Does the new Codex update only affect coding tasks?

No. OpenAI explicitly framed the release as support across the software development lifecycle and into repeatable workflow tasks beyond pure coding.

Is computer use available everywhere?

Not yet. OpenAI said computer use is initially available on macOS and will roll out to EU and UK users later.

Why should ClawNewbie readers care about this launch?

Because it changes how Codex should be compared with other coding tools. The product is now being positioned more as a workflow agent and less as a narrow coding assistant.

• OpenAI, Codex for (almost) everything, published April 16, 2026, checked April 23, 2026: https://openai.com/index/codex-for-almost-everything/