Skip to content
News Product launch

OpenAI Codex Remote reaches GA with QR pairing and a DigitalOcean Droplet plugin

· by Pondero Newsdesk

The short version

OpenAI shipped Codex Remote GA on June 25, replacing its previous connection model with one-to-one QR pairing and adding a DigitalOcean Droplet plugin for persistent cloud workspaces.

OpenAI Codex Remote reaches GA with QR pairing and a DigitalOcean Droplet plugin

OpenAI moved Codex Remote to general availability on June 25, 2026, extending phone-based control of long-running coding sessions to all paid ChatGPT plans and shipping a new DigitalOcean Droplet plugin that lets Codex provision and attach a cloud server as a persistent remote workspace.

What changed

The GA release covers every paid ChatGPT tier: Plus, Pro, Business, Enterprise, and Education, per OpenAI's release notes. The feature lets developers start or continue Codex work on a connected Mac or Windows host from the ChatGPT mobile app, review agent progress, and approve actions without sitting at the machine.

The biggest security change is the pairing model. The previous remote connection required a more loosely bound link between host and phone. GA replaces that with authenticated one-to-one QR pairing: each mobile device must be paired to each host individually, with the pairing validated against the account's existing authentication chain including MFA, SSO, or passkey steps. Connections active since June 8 carry over after updating both the Codex desktop app and the ChatGPT mobile app. Connections dormant before June 8 need a fresh pair. Signing out of ChatGPT disables remote control but does not delete existing pairings, so they restore on sign-in.

The relay architecture keeps host machines off the public internet. The Codex app on the host communicates outbound through OpenAI's relay, which syncs session state to any authorized phone without opening inbound ports. Projects, credentials, plugins, and MCP servers stay local; the relay passes only session messages (prompts, approvals, diffs, terminal output) between the devices.

The DigitalOcean Droplet plugin

Alongside the Remote GA, OpenAI released a DigitalOcean Droplet Workspace plugin for Codex. The plugin lets Codex provision a Droplet in the user's DigitalOcean account via natural language, configure SSH access to it, and attach it as a persistent remote workspace. Tasks running on a Droplet are unaffected by local machine sleep or network drops. The Droplet costs standard DigitalOcean pricing billed to the user's own account; the plugin itself ships as part of Codex.

For developers who want to route heavy or long-running Codex work through a cloud machine without managing separate infrastructure, DigitalOcean becomes a supported option directly from within Codex rather than a manual side step.

Why it matters

Two things make this more than a feature flag flip to GA.

First, the security timing is pointed. In May 2026, Aikido Security researchers documented a supply chain attack where a malicious npm package posing as a Codex remote UI reached more than 29,000 weekly downloads while exfiltrating developer tokens to an attacker-controlled server, as TechTimes reported. A stolen Codex refresh token grants persistent account access without a password. Shipping a relay architecture that opens no inbound ports, plus one-to-one QR binding, directly addresses the credential-exposure surface that attack exploited.

Second, the Droplet plugin fits a pattern OpenAI has been building out. On June 11, 2026, OpenAI announced it acquired Ona, formerly Gitpod, whose platform runs AI coding agents in persistent cloud sandboxes. The DigitalOcean plugin, in that context, looks like one piece of a broader move toward agents that continue working after a developer closes their laptop, not just agents that can be supervised from a phone.

For developers already on a paid ChatGPT plan, Codex Remote is now available without a waitlist. The practical workflow question is whether approving and steering agent work from a phone changes how teams structure long-running tasks. The relay model makes it frictionless enough to try.

What to watch next

Watch whether OpenAI ships additional cloud-provider plugins in the DigitalOcean Droplet pattern. AWS, GCP, and Azure are the obvious next targets, and the Ona acquisition suggests the team has the infrastructure thinking in place. Also watch for a DigitalOcean-published integration guide, which would formalize the pairing as a supported two-vendor workflow rather than a plugin OpenAI ships unilaterally.

Sources