Codex v26.527 on Windows: Computer Use, Mobile Remote, and How It Stacks Up Against macOS

2. Remote control from the ChatGPT mobile app

Your phone is now a remote control for your Windows machine. You can initiate Codex tasks on your PC directly from the ChatGPT mobile app on iOS or Android, or from another Mac running Codex, and monitor the agent's progress in real time . This is more than a simple notification system—the mobile experience surfaces the live state of the environment, including project context, active threads, pending approvals, terminal output, diffs, and screenshots . You can approve an action, answer a clarifying question, or redirect a task while waiting for your coffee. This feature is designed to keep longer-running investigations and complex multi-step projects moving, even when you are not at your desk .

3. New Profile section

The update also adds a Profile section to the Windows app, giving users a cleaner interface to view profile details, track usage stats, and monitor token activity .

Windows vs. macOS: A direct comparison

With this release, Codex on Windows closes much of the capability gap, but a few important differences remain. Most core features are now consistent across both platforms , but power users will notice a distinction in how "computer use" is implemented.

The critical difference is locked and background computer use. On macOS, eligible users can enable "locked computer use," which allows Codex to continue working securely in the background after the Mac's screen is locked. This is paired with a strict security model involving short-lived authorizations and display safeguards . In practice, you can start a multi-hour migration test, lock your MacBook, and check its progress from your phone later. On Windows, Computer Use currently operates exclusively in the foreground while the machine is actively in use . OpenAI has not indicated if or when a locked computer use mode will arrive for Windows.

Another macOS-exclusive is Appshots. This feature allows users to attach a screenshot of any application window to a Codex thread using a hotkey, providing the agent with the visual layout and any available text for context . The feature has not been mentioned in any Windows documentation or changelogs for v26.527.

Other advanced features, including Goal mode, in-app browser annotations, automations, and native sandbox execution via PowerShell, are available across both Windows and macOS .

What this means for your workflow

The shift is philosophical. A "code generation agent" is a tool for developers. A system that clicks and types in any application is a tool for everyone. For Windows users specifically, this update removes a major platform asterisk. The workflow of starting a complex research task on your work PC and then steering it from your phone on the way home is no longer a macOS-only perk.

The main question for Windows professionals is how much the absence of a locked computer use mode limits real-world productivity. If your workflows are fully interactive and require a human in the loop, the foreground-only operation on Windows may not be a blocker. But for users who rely on long-running, unattended background automations, macOS remains the more capable platform for now. Whether future Windows updates add this capability remains an open question, as the current documentation makes no mention of it .