OpenAI added Codex to mobile ChatGPT for managing coding tasks from anywhere
OpenAI brought Codex to mobile ChatGPT. Developers can now use their phones to track coding tasks, steer them as they run, and confirm actions in real time, eve

OpenAI has expanded access to Codex and embedded it into the mobile ChatGPT application. Now developers can track code tasks, direct them, and confirm actions in real time even away from their laptop.
Codex on Your Phone
The main idea behind the announcement is simple: Codex is no longer tied to the desktop. If a task is already running on another device or in a remote environment, a user can open ChatGPT on their smartphone and quickly check its status, see the progress, and intervene if necessary. This is particularly useful when code is being generated, tests are running, and the solution requires human confirmation before the next step. Mobile access transforms Codex from a tool for a computer session into a constant work channel.
According to OpenAI's description, it's not just about viewing status. Through mobile ChatGPT, you can provide further instructions, direct the course of tasks, and confirm actions as needed. In effect, the company is betting on a scenario where the developer remains in the decision-making loop, but is no longer obligated to sit in front of the screen all the time. This format is especially appropriate where work has already been moved to a remote environment or a separate device.
How the Workflow Looks
The new mode appears to be a logical continuation of the idea of asynchronous development with an AI assistant. You launch a task where it's convenient to work with the repository and environment, and then monitor it from your phone while heading to a meeting, traveling across the city, or simply stepping away from your desk. If Codex hits a fork, asks for confirmation, or shows an intermediate result, you don't need to return to your laptop just for a single click.
In this format, the mobile application covers several practical scenarios:
- Checking task status and progress of execution
- Quick confirmation of actions that require human participation
- Adjusting course if you need to clarify a request or change the plan
- Controlling tasks launched on another device or in a remote environment
"Monitor, direct, and confirm coding tasks in real time."
This is not an attempt to replace a full-fledged IDE on a smartphone. Rather, it's about a mobile control panel for those moments when work is already underway and you need to control the process rather than write code from scratch. This approach fits well with the modern stack, where computing, agents, and environments increasingly live separately from the developer's local device, and a decision needs to be made without being tied to a single screen.
Why This Is Convenient
For OpenAI, this is another step toward a model in which Codex becomes not a one-time function, but a constant agent alongside the developer. The less friction between a task, a device, and the moment of decision-making, the easier it is to embed AI into the actual workday. If previously many scenarios ended the moment a person stepped away from the computer, now the process gains continuity: the task continues to live, and the user remains connected.
For teams, this is also a practical change. When tasks are executed in remote environments, on servers, or on another device, mobile access reduces the delay in coordination and unlocks the next step faster. There's no need to wait for a person to return to their workstation to confirm an action or correct an instruction. If OpenAI can bring this scenario to stable everyday use, the boundary between "I am coding" and "I am managing coding" will become even thinner.
What This Means
OpenAI is moving Codex toward an always-available assistant rather than a tool only for desktop sessions. For developers, this signals that managing AI tasks is gradually becoming multi-platform and asynchronous: part of the work can be launched in one place and controlled and approved in another, without losing momentum. If this mode takes hold, mobile ChatGPT will become a working interface not only for conversation, but also for daily development support.