When it comes to openai codex now control, understanding the latest developments is essential. This post contains affiliate links. We may earn a commission at no extra cost to you.
I spent time digging into this one, and here’s my honest take.
TOOL UPDATE
OpenAI Codex
“For (Almost) Everything”
Computer UseControl your Mac
MemoryLearn your workflow
90+ PluginsExtend everything
toolsstackai.com • April 2026
OpenAI just made Codex a LOT more than a coding assistant. The April 16 update — cheekily titled “Codex for (almost) everything” — adds computer use, an in-app browser, persistent memory, image generation, and over 90 new plugins. For the 3+ million developers already using Codex, this is the biggest update since launch.
I’ve been following the AI coding assistant space closely, and this update caught my attention because it shifts Codex from “smart autocomplete” to something closer to a full-fledged AI coworker that can actually operate your computer. Let me break down what’s new and whether it lives up to the hype.
Computer Use: Codex Can Now Control Your Mac
This is the headline feature, and it’s a big one. Codex can now operate desktop Mac apps using its own cursor — seeing what’s on screen, clicking, and typing to complete tasks. Think of it as giving Codex hands. Need it to test your app in a browser? Configure a database GUI? Run a sequence of actions across multiple desktop apps? It can do that now.
The really clever part: Codex can run multiple agents in parallel on your Mac without interfering with your own work. So while you’re writing code in one window, Codex could be running tests, checking documentation, and reviewing PRs in separate processes simultaneously.
In-App Browser: Browse, Annotate, and Direct
Codex now includes a built-in browser where you can navigate to any page and leave comments directly on elements to give the AI precise instructions. See a bug on your staging site? Open it in the Codex browser, click on the broken element, and type “fix this alignment.” That’s the workflow they’re going for.
OpenAI plans to expand this so Codex can fully command the browser beyond just localhost web applications, which would make it capable of interacting with web services, APIs, and external tools directly.
Memory: It Finally Remembers Who You Are
This one might sound simple, but it’s a massive quality-of-life improvement. Codex now remembers your preferences, recurring workflows, tech stacks, and personal patterns. No more explaining your project setup every time you start a new session.
Even more useful: Codex can now resume work after a pause using existing conversation threads and even schedule future work for itself across days or weeks. Start a refactoring project on Monday, and Codex can pick it back up on Wednesday exactly where it left off.
90+ New Plugins: The Ecosystem Explodes
OpenAI added over 90 plugins that combine skills, app integrations, and MCP (Model Context Protocol) servers. This is Codex’s version of an app store, and it significantly extends what the tool can do beyond pure coding.
| Feature | What’s New | Status |
|---|---|---|
| Computer Use | Control Mac apps, parallel agents, screen reading | Live (macOS only) |
| In-App Browser | Browse pages, annotate elements, direct AI actions | Live |
| Memory | Persistent preferences, workflow continuity, self-scheduling | Live |
| Image Generation | GPT-Image-1.5 for in-app visuals | Live |
| 90+ Plugins | Skills, integrations, MCP servers | Live |
| Developer Tools | PR reviews, multi-file view, SSH remotes, terminals | Live |
How This Stacks Up Against the Competition
The AI coding space is heating up fast. Cursor just launched version 3.0 with parallel AI agent fleets and its Composer 2 model. Anthropic’s Claude Code has been gaining ground with developers who prefer its reasoning capabilities. GitHub Copilot continues to evolve. And smaller players like Lindy AI are carving out niches in AI agent building.
What sets Codex apart with this update is the computer use capability. While other tools are confined to the code editor, Codex can now operate across your entire desktop. That’s a fundamentally different value proposition — it’s not just helping you write code, it’s helping you do everything around the code too.
For developers who also handle deployment, testing, documentation, and project management (which is… most developers), having an AI that can operate desktop apps is a genuine productivity multiplier.
Should You Switch to Codex (Or Switch Back)?
If you left Codex for Cursor or another tool, this update is worth a fresh look. The computer use and memory features address two of the biggest complaints developers had: context loss between sessions and the inability to do anything outside the editor.
If you’re already using Codex, the update should roll out automatically. Check for the new features in your app settings.
For developers building AI-powered applications, combining Codex with automation tools like Zapier or n8n could create powerful workflows — Codex handles the code and desktop tasks while automation platforms manage the broader workflow orchestration.
Quick FAQ
Is Codex computer use available on Windows?
Not yet. Computer use launched as macOS-only on April 16, 2026. No timeline for Windows or Linux support has been announced.
Do I need to pay extra for these features?
The features are included in existing Codex subscriptions. Plugin availability may vary by plan tier.
Can Codex run multiple tasks simultaneously?
Yes. Codex can run multiple parallel agents on your Mac without interfering with your own work, handling tasks like testing, PR reviews, and documentation simultaneously.
How does Codex memory work?
Codex learns your preferences, tech stack, recurring workflows, and personal patterns over time. It can also resume paused work and schedule future tasks for itself.
Is this better than Cursor 3.0?
They serve different strengths. Codex offers broader capabilities (desktop control, browser, plugins) while Cursor 3.0 excels at deep, parallel codebase work. Many developers use both.
