OpenAI Brings Codex to the ChatGPT Mobile App
OpenAI's Codex coding agent now runs through the ChatGPT mobile app, letting developers monitor, steer and approve coding tasks in real time from any device or remote environment.

Updated
Why it matters
- Codex is now available through the ChatGPT mobile app.
- Users can monitor, steer and approve coding tasks in real time.
- The capability works across devices and remote environments.
OpenAI has brought Codex to the ChatGPT mobile app, letting developers work with the coding agent from anywhere. The company announced the change in a post titled "Work with Codex from anywhere."
The feature covers three core actions: monitoring, steering, and approving coding tasks. Users can watch an agent's progress in real time, redirect its work mid-task, and sign off on completed steps — all from a phone or tablet, without sitting at a development machine.
The capability works across devices and remote environments. That means a task started on a desktop can be supervised from a phone, and work running in a remote environment stays reachable through the same mobile interface.
The move matters because it extends Codex beyond the terminal and IDE. Agentic coding tools increasingly run long, multi-step tasks that finish while the developer is away from the keyboard. Mobile supervision turns dead time — a commute, a meeting, a lunch break — into moments where an agent's output can be reviewed and unblocked.
For teams running agents against shared or remote environments, approval workflows on mobile also reduce bottlenecks. A task that needs a human sign-off no longer waits until that person returns to their desk.
OpenAI is competing in a crowded agentic-coding market where the ability to supervise work from anywhere is becoming table stakes. Bringing Codex into the ChatGPT mobile app also tightens the integration between the coding agent and the company's main consumer surface, rather than shipping it as a separate standalone product.
Expect the gap between starting a coding task and finishing it to keep shrinking: if approval and steering live in every developer's pocket, the agent can keep working while the human decides when to look.
Source: OpenAI News
More from James Calloway
Show full bio
News editor covering industry trends and analytics at AI In Context.
119 articles