Access Codex remotely from mobile apps
TL;DR
Codex is now available in preview on iOS and Android, allowing users to connect to a Mac host. You can review terminal output, approve actions, and manage active investigations while away from your desktop.
What changed
Codex remote access entered preview on May 14, 2026. Users can now connect the ChatGPT iOS or Android app to a Mac host running the Codex desktop app. From the phone they can start threads, approve actions, review terminal output, diffs, screenshots, and test results, then switch hosts without returning to the desk.
The host must stay awake and online. Setup begins in the Codex app on macOS, continues by scanning a QR code in ChatGPT, and works across Free, Go, Plus, and Pro plans in supported regions.
Why it matters
Vibe Builders who keep Codex running long investigations or agent loops gain the ability to steer work from anywhere. The change reduces the friction of being tied to one machine during multi-hour tasks.
It pressures desktop-only automation tools that lack mobile oversight. The bet is that builders will accept always-on Mac hosts in exchange for continuous control rather than batching work until they return to their desk.
How to use it
Update the ChatGPT mobile app and the Codex macOS app. Open the Codex app on the host, enable remote connections, then scan the QR code shown inside ChatGPT. Keep the Mac awake and connected for the session to persist.
Remote access loads live project context, approvals, plugins, and terminal output directly from the host. Documentation sits at developers.openai.com/codex/remote-connections.
Watch for
Confirm the bet if mobile sessions show stable context and low latency over several hours. It breaks if hosts drop connections frequently or if approval flows feel slower than desktop. Expect OpenAI to add Windows hosts or background push notifications next.
Who this matters for
- Vibe Builders: Monitor long-running agent loops and approve terminal actions via the ChatGPT mobile app.
- Basic Users: Connect your phone to your Mac via QR code to check on active AI tasks while away from your desk.
Harsh’s take
Mobile oversight for Codex is a major win for operators running persistent agentic loops. The requirement for an always-on Mac host is a fair trade for the ability to approve terminal commands or review diffs from a phone. This feature effectively kills the dead time between starting a complex task and returning to a workstation to verify the result.
OpenAI is correctly identifying that the bottleneck for builders is not the compute, but the human-in-the-loop latency. By moving approvals to mobile, they keep the momentum of the build alive. This puts immediate pressure on desktop-only IDE extensions that lack a robust mobile companion.
If you are running multi-hour investigations, this setup is now the baseline for maintaining high velocity without being tethered to a physical desk.
by Harsh Desai
About ChatGPT
View the full ChatGPT page →All ChatGPT updatesGo deeper
More AI news
- FeatureHermes Agent verifies work with completion contracts and evidence ledgers
Hermes Agent records verification evidence for coding tasks. The /goal command uses completion contracts to judge success against test runs rather than model assertions.
- FeatureCursor adds cloud agent management to the Agents window
Cursor sets up cloud development environments in under 10 minutes, spins up isolated cloud subagents using /in-cloud, and hands off sessions between local and cloud.
- FeatureCursor introduces /automate skill for automating repetitive tasks
Cursor's new /automate skill creates automations from plain language. Workflows trigger via Slack emojis or GitHub events while cloud agents access virtual computers.