OpenAI Codex goes mobile with ChatGPT app integration for remote coding workflows
OpenAI Codex now runs inside the ChatGPT iOS and Android apps, letting developers monitor and manage workflows remotely from their phones.
OpenAI on Thursday integrated OpenAI Codex into the mobile ChatGPT app, enabling developers to view and control Codex live environments from iOS and Android devices. The preview release, available to all subscription tiers, lets users monitor running agents, dispatch tasks, approve outputs and switch models without returning to a desktop. OpenAI said the update extends Codex’s reach beyond a single computer and aims to make agentic coding more practical for on-the-go teams.
How the mobile integration works
OpenAI Codex appears inside the ChatGPT app as a panel where users can see active Codex threads and live execution environments.
From a phone or tablet, developers can inspect logs, review intermediate outputs, and approve or halt commands that Codex executes on remote machines.
The integration also supports model changes and the initiation of new tasks, meaning managers can adjust workflows in real time during meetings or while away from their desks.
Capabilities available on mobile
Users can remotely approve critical steps in multi-stage automation, rerun failed commands, or adjust parameters across sessions.
The mobile interface surfaces essential context such as recent commits, execution timestamps, and brief summaries of agent activity to help rapid decision-making.
OpenAI emphasized that this is not merely remote control of a single job, but a way to move across all threads, review outputs, and start new workstreams from a handheld device.
Recent feature expansions that set the stage
This mobile update follows months of rapid enhancements to Codex’s autonomy and reach.
Last month, OpenAI gave Codex the ability to run persistently in desktop environments so agents can perform background tasks autonomously. The company also released a browser extension that allows Codex to operate in live Chrome sessions, bridging local development and web-based workflows.
Together these releases position Codex as a cross-environment agent that can originate work in a browser, perform sustained tasks on a workstation, and now be supervised from a mobile device.
Competitive pressure from Anthropic
Anthropic introduced a similar capability, Remote Control for Claude Code, earlier this year, enabling remote monitoring of agentic coding work.
The arrival of mobile supervision features from multiple vendors reflects an emerging market contest over which agentic coding tool becomes the default for enterprises and professional developers.
Industry observers say the rivalry is driving faster rollouts and feature parity as both companies seek adoption among teams that value continuous automation and oversight.
Enterprise and developer use cases
Product and engineering managers can now approve deployments or inspect long-running experiments while commuting or between meetings.
Remote teams can hand off monitoring duties without requiring VPN access to a particular workstation, which may simplify on-call rotations and incident response.
OpenAI positions the feature as helpful not only for individual developers but also for stakeholders who need visibility into automated pipelines without deep technical setup.
Availability, privacy and safety notes
The feature is rolling out in preview to all ChatGPT plans on iOS and Android, according to OpenAI’s announcement.
OpenAI said the mobile controls allow changes to models and commands, but it also reiterated earlier commitments to guardrails and oversight when agents act autonomously. The company pointed to mechanisms for user approval and the ability to halt or modify commands in real time.
Enterprises weighing the mobile preview should evaluate access controls, audit logging, and network security to ensure remote supervision fits their compliance posture.
OpenAI’s Codex integration into the ChatGPT mobile app marks a notable step in making agentic coding more portable and responsive to modern workflows. The update narrows the gap between where code runs and where decisions about that code are made, putting real-time supervision and lightweight management in the palm of developers’ hands.
