Which model does the Codex CLI or IDE extension use?

Codex primarily utilizes the GPT-5.1-Codex model family, with 'Max' as the default and 'Mini' as an optional choice. It's crucial to note that GPT-4o is not available within Codex, and older models like GPT-4o cannot be restored or purchased as legacy tiers. Users can select models via the model picker, including legacy options, or specify them using the '-m' flag or in the 'config.toml' file. Model availability and pricing for the ChatGPT platform and the API are managed separately from Codex, and users should consult the Codex documentation for current supported models and configuration guidelines. This ensures developers can leverage the most effective models for their coding tasks while understanding the distinct model ecosystems.

Does OpenAI train on my Codex usage?

For business users, including those with ChatGPT Business, Enterprise, and API plans, OpenAI explicitly states that inputs and outputs from their products are not used by default to train or improve their models. However, API organization owners have the option to opt-in and share API data, though this setting is unavailable for Enterprise and Zero Data Retention accounts. Conversely, for Pro and Plus users, conversations may be utilized for model improvement unless the user actively disables this feature through their ChatGPT data controls. Users are encouraged to review OpenAI's documentation on data usage to understand how their information contributes to model performance and to manage their privacy settings accordingly.

Can I run Codex in my IDE?

Yes, developers can integrate and run Codex directly within their Integrated Development Environments. The dedicated Codex VS Code extension is designed for compatibility with most VS Code forks, offering a streamlined experience for code generation, review, and task delegation. For users of other IDEs that may not have a native Codex extension, the Codex Command Line Interface (CLI) provides a versatile alternative. The CLI can be executed within the IDE's integrated terminal, allowing developers to access Codex functionalities and leverage its AI coding capabilities without needing to switch environments. This flexibility ensures a wide range of development setups can benefit from Codex.

Are there separate workspace permissions needed for the Codex app?

No, the Codex app operates under the same unified workspace controls that govern all other Codex interfaces and functionalities. This means that if you have established permissions for Codex Local (which covers usage through the CLI and IDE extension) and Codex Cloud (which dictates whether team members can delegate tasks from the Codex app and other cloud-based interfaces), these permissions will apply seamlessly to the Codex app. There's no need to configure additional, distinct permissions specifically for the app. For a comprehensive understanding of setting up and managing workspace permissions, users should refer to the Enterprise Admin Guide for Codex, ensuring consistent access management across all deployment surfaces.

How can I access Codex Enterprise Analytics?

Access to Codex Enterprise Analytics is exclusively provided to Enterprise workspaces that have Codex enabled as part of their suite of tools. If your organization operates within an Enterprise workspace and wishes to leverage these analytics, the primary step is to ensure that Codex Cloud functionalities are activated for your organization. This typically requires liaison with your workspace administrator. Once enabled, detailed setup instructions and API access points for retrieving analytics data can be found within the comprehensive API documentation provided by OpenAI. This ensures administrators can monitor usage, performance, and other key metrics relevant to their enterprise-level deployment of Codex.

Where can I find more info for troubleshooting Codex app issues?

For users encountering issues or requiring assistance with the Codex app, OpenAI provides dedicated resources to aid in troubleshooting. The primary reference point is the comprehensive platform guide available on the OpenAI developer website, specifically tailored for the Codex app troubleshooting. This guide contains detailed information, common solutions, and diagnostic steps to help resolve various problems. Should the platform guide not address a specific or complex issue, users are encouraged to directly contact OpenAI's support team. This ensures that expert assistance is available to help maintain smooth and efficient operation of the Codex app within your development workflow.

Codex Integration: Revolutionizing Your ChatGPT-Powered Development Workflow

In the rapidly evolving landscape of software development, AI tools are becoming indispensable. OpenAI's Codex stands out as a powerful AI coding agent designed to streamline and accelerate the coding process. Seamlessly integrated with various ChatGPT plans, Codex empowers developers to write, review, and deploy code with unprecedented efficiency. This article delves into how developers can leverage Codex within their existing ChatGPT subscriptions, covering everything from initial setup to advanced enterprise features and usage considerations.

It's important to clarify that while OpenAI offers a suite of advanced AI capabilities, Codex specifically focuses on coding tasks. Features like image and video generation (Sora, ChatGPT Images, DALL·E) and Voice features operate under separate usage limits and reset periods. If you encounter banners related to these features, rest assured that those restrictions do not impact your Codex usage.

Streamlining Development with OpenAI Codex Integration

Codex is more than just a code generator; it's an intelligent AI coding agent built to function as a collaborative partner in your development cycle. Whether you're a solo developer or part of a large enterprise team, Codex helps you navigate repositories, edit files, execute commands, and run tests, significantly boosting productivity. It's designed to integrate deeply into your daily workflow, offering flexible interaction methods to suit diverse preferences and environments.

For a limited time, Codex is broadly accessible, included with ChatGPT Plus, Pro, Business, and Enterprise/Edu plans. During this promotional period, even ChatGPT Free and Go users can explore Codex, while other plans benefit from 2x rate limits, maximizing their AI-powered coding potential.

Unleash AI Power: Pairing and Delegating with Codex

Codex offers versatile modes of operation, allowing developers to choose how they interact with the AI coding agent for optimal productivity.

Pairing in Your Local Development Environment

You can pair directly with Codex in your preferred local tools. Starting from a simple prompt or a detailed specification, Codex can intelligently navigate your repository to make edits, execute commands, and run tests. This hands-on integration is available through:

Codex CLI: A command-line interface for direct interaction.
Codex IDE Extension: Available for popular environments like VSCode, Cursor, and Windsurf, bringing AI assistance directly into your editor. Developers increasingly rely on extensions for "agentic coding" workflows, which Codex greatly facilitates, akin to capabilities explored in articles like Xcode Agentic Coding.
Codex App: For those managing multiple projects or seeking a unified interface, the Codex app offers built-in worktree support, skills, automations, and Git functionality, eliminating the need to switch between terminals or IDEs. The app is available for macOS and Windows.

Delegating Tasks to Codex in the Cloud

For more extensive or background tasks, Codex can operate in the cloud, running in an isolated sandbox with your repository and environment. This allows you to stay focused on your primary work while Codex handles tasks in the background, generating code that you can later review, merge, or pull down for further local development. This cloud delegation is particularly useful for complex operations or when you need to offload computational tasks.

Automating Code Review with Codex

One of Codex's most impactful features is its ability to automate code reviews. Just like a human colleague, Codex can analyze your pull requests (PRs) directly within GitHub. You have the flexibility to enable automatic reviews for your personal PRs or configure it for your entire team across multiple repositories, fostering higher code quality and accelerating the review process. This capability aligns with the growing trend of GitHub Agentic Workflows, where AI assists in various stages of the development lifecycle.

Seamlessly Integrating Codex with Your ChatGPT Account

Getting started with Codex using your existing ChatGPT plan is straightforward, ensuring a smooth transition into AI-assisted development.

To connect Codex with your Plus, Pro, Business, Edu, or Enterprise plan:

Ensure an active ChatGPT subscription. If you're not subscribed, you can easily upgrade via your ChatGPT account settings.
Launch your preferred Codex client (CLI, IDE extension, web interface, or desktop app).
Follow the client's instructions to sign in with your ChatGPT account.

For the Codex web interface, you will need to connect your ChatGPT account to your GitHub account.
For programmatic control, the Codex SDK is available, and integrations like asking Codex to perform tasks via Slack are also supported.

Note for existing API key users: If you've previously accessed the Codex CLI via an API key, remember to update your package manager, then run codex logout followed by codex to switch to subscription-based access and take advantage of the integrated ChatGPT plans. For in-depth details on features and usage, the comprehensive Codex documentation page is an invaluable resource.

Enterprise-Grade Deployment and Control for Codex

For larger organizations, Codex offers robust features designed for enterprise-level deployment, management, and compliance, ensuring secure and scalable integration within corporate environments.

Enterprise Setup and Role-Based Access Control (RBAC)

Setting up Codex for an enterprise workspace involves detailed configuration covered in the Enterprise Admin Guide. This guide provides a comprehensive roadmap for getting your workspace fully operational. Furthermore, access to Codex can be finely tuned using Role-Based Access Control (RBAC), allowing administrators to grant specific user roles appropriate permissions. More information on establishing RBAC can be found in the dedicated RBAC guide.

Data Residency and Compliance API

Organizations prioritize data governance, and Codex is designed with this in mind. For Codex usage when delegated to the cloud or via the web, activity logs are available through the Compliance API. It's important to note that usage in local environments (CLI or IDE extension) is not recorded via this API. Codex ensures compliance with stringent Data Retention & Residency policies, a critical aspect for Enterprise Privacy considerations.

Understanding Codex Usage Limits and Optimizing Performance

The utility of Codex, like many AI services, is subject to usage limits that vary based on your specific ChatGPT plan.

Codex Usage Limits by Plan

Plan Type (ChatGPT)	Codex Inclusion	Rate Limits	Factors Influencing Usage
Free & Go (Limited Time)	Included	Standard	Task size, code complexity, context required, execution environment (local vs. cloud).
Plus, Pro, Business, Edu, Enterprise	Included	2x Rate Limits (Limited Time)	Task size, code complexity, context required, execution environment (local vs. cloud). Larger codebases or long-running tasks consume significantly more allowance.

During a limited promotional period, Plus, Pro, Business, and Enterprise/Edu subscribers benefit from 2x Codex rate limits. The number of Codex messages you can send within these limits is not fixed but rather dynamically consumed. Factors such as the size and complexity of your coding tasks, and whether tasks are executed locally or in the cloud, heavily influence usage. For instance, small scripts or simple functions might consume only a fraction of your allowance, whereas processing larger codebases, handling long-running tasks, or extended sessions requiring Codex to maintain significant context will utilize substantially more. For a comprehensive breakdown of Codex limits and pricing, developers are encouraged to consult the pricing page on the OpenAI developer website.

By understanding these dynamics and utilizing Codex efficiently, developers can maximize the benefits of this powerful AI coding agent, truly enhancing their software development lifecycle.

Codex Integration: Enhancing Your ChatGPT Coding Workflow