OpenAI Codex
codingby openai
OpenAI's multi-surface coding agent spanning a desktop app, CLI, IDE extension, cloud, and mobile remote control.
Key features
Free tier available, Plus at $20/mo
Developers who want a multi-surface coding agent that works across desktop, terminal, IDE, cloud, and mobile from a single configuration
True multi-surface product: desktop app, CLI, IDE extension, cloud, and mobile remote control all share one configuration, so you can start work on one surface and continue on another.
What it does
Five Product Surfaces
Available as a desktop app, open source CLI, IDE extension, web/cloud interface, and through the ChatGPT mobile app. All five share the same agent, configuration, and MCP servers.
Learn moreCloud Sandbox Execution
Codex Cloud runs each task in its own isolated sandbox environment, preloaded with your repository. Your local machine stays untouched while Codex works in the background and delivers pull requests.
Learn moreAutonomous Task Completion
Describe what you need in plain English and Codex handles the rest: reading your code, planning changes, writing new code, and running tests to verify everything works.
OS Native Sandboxing
Uses operating system level sandboxing by default (Seatbelt on macOS, bubblewrap on Linux, Windows Sandbox on Windows), providing stronger isolation than permission prompts alone.
Learn moreComputer Use
The desktop app can operate GUI applications by seeing, clicking, and typing. Useful for testing desktop apps, checking browser flows, reproducing bugs, and interacting with data sources that have no plugin or connector.
Automations
Schedule recurring agent tasks in the desktop app. Standalone automations run on a cron schedule with results in a Triage inbox. Thread automations provide heartbeat style wake-ups that preserve conversation context across days or weeks.
Learn moreMobile Remote Control
Control Codex from your phone via the ChatGPT mobile app. Start new tasks, steer active work, approve commands, review diffs and screenshots, and switch between connected host machines, all through a secure relay.
Learn moreCodex CLI (Open Source)
An open source terminal agent built in Rust, licensed under Apache 2.0. Runs locally on your machine, supports multiple AI providers, and features three sandbox modes with OS native enforcement.
Learn moreMCP Client and Server
Codex connects to third-party MCP servers (via STDIO and Streamable HTTP) and can also run as an MCP server itself, letting other agents invoke it for tasks like code review or delegation.
IDE Extension with Cloud Delegation
Available for VS Code, Cursor, Windsurf, and JetBrains IDEs. Supports offloading tasks to Codex Cloud from within the editor, preserving conversation context, and pulling results back locally.
Parallel Task Execution
Run multiple Codex tasks at the same time. Each one gets its own sandbox, so you can tackle several parts of a project simultaneously in the cloud or across local threads with git worktree isolation.
GitHub Integration
Codex connects directly to your GitHub repositories. Tag @codex on GitHub issues or pull requests to spin up cloud tasks. Completed work is delivered as pull requests for review.
Appshots
Press both Command keys on macOS to capture the frontmost app window (screenshot plus available text) and send it to a Codex thread. This is how Codex gets context from non-code apps like email, design tools, or error dialogs.
AGENTS.md Configuration
A project level configuration file that tells Codex your conventions, setup commands, and testing requirements. Similar to CLAUDE.md for Claude Code. Supports global and per-directory placement.
Learn moreCodex Security Agent
A dedicated security agent launched in March 2026 that autonomously scans your codebase for vulnerabilities, identifies security issues, and generates fixes delivered as pull requests.
Figma Integration
Connect Figma designs directly to Codex. It reads your design files and translates them into code, bridging the gap between design and implementation workflows.
Pricing
Free
Basic Codex access for exploration. Included with the free ChatGPT tier. Limited usage for getting started with Codex CLI and basic tasks.
- Codex CLI access
- Basic exploration and learning
- Limited usage per 5-hour window
- Community support
Go
Use Codex for lightweight coding tasks. A step up from Free with more access, but without the full model suite or cloud tasks available on Plus.
- More Codex usage than Free
- Lightweight coding tasks
- Codex CLI access
Plus
Focused coding sessions with access to GPT-5.5, GPT-5.4, GPT-5.3-Codex, and GPT-5.4-mini. Includes Codex Cloud tasks and the full desktop app experience.
- GPT-5.5, GPT-5.4, GPT-5.3-Codex, GPT-5.4-mini
- Codex Cloud (parallel background tasks)
- Desktop app with Computer Use and Automations
- Mobile remote control via ChatGPT app
- GitHub pull request creation
- IDE Extension with cloud delegation
Pro 5x
5x the usage limits of Plus. Designed for developers who use Codex frequently and need more messages per 5-hour window across all models.
- Everything in Plus
- 5x usage limits across all models
- GPT-5.3-Codex-Spark (research preview)
- Priority execution
- Higher cloud task limits
Pro 20x
20x the usage limits of Plus. The highest individual tier for power users who rely on Codex as their primary development environment.
- Everything in Pro 5x
- 20x usage limits across all models
- Maximum cloud task throughput
- Extended context for larger codebases
Business
Pay-as-you-go usage pricing for teams with no fixed seat fee. Includes a secure, dedicated workspace with essential admin controls, SAML SSO, and MFA.
- Everything in Plus
- No fixed seat fee, pay as you go based on usage
- Larger virtual machines for faster cloud tasks
- Flexibly extend usage with ChatGPT credits
- Secure workspace with SAML SSO and MFA
- No training on your business data by default
Pros & Cons
Pros
- True multi-surface product: desktop app, CLI, IDE extension, cloud, and mobile remote control all share one configuration, so you can start work on one surface and continue on another.
- OS native sandboxing by default (Seatbelt, bubblewrap, Windows Sandbox) provides stronger isolation than permission prompts alone, reducing the risk of accidental damage.
- The CLI is fully open source under Apache 2.0, built in Rust, with over 75,000 GitHub stars. Supports multiple AI providers beyond OpenAI.
- Mobile remote control via the ChatGPT app is a genuine cross-device capability: start tasks, approve commands, review diffs, and steer work from your phone through a secure relay.
- Automations allow scheduling recurring agent tasks on a cron schedule, filling a niche that most competing tools require external cron jobs or third-party orchestration to handle.
- Has a free tier to start with, and the CLI is free to use with any compatible API key. Lower barrier to entry than tools that require a paid subscription from day one.
- MCP support in both directions: connects to third-party MCP servers as a client and can run as an MCP server itself, letting other agents (including Claude Code) invoke it.
Cons
- Cloud execution means no access to local tools, databases, or services. If your project depends on a local database or API for testing, Codex Cloud cannot reach it (though the CLI and desktop app can).
- Computer Use is not available in the EEA, UK, or Switzerland at launch. This is a significant geographic restriction for European users.
- The desktop app and IDE extension are not open source. Only the CLI, SDK, and App Server are under the Apache 2.0 license.
- Less extensible than Claude Code overall. While Codex has hooks, skills, MCP, and Automations, it lacks the depth of Claude Code's four-tier CLAUDE.md hierarchy, 27 hook events, managed enterprise MCP allowlists, and markdown-as-infrastructure customization system.
- GPT-5.5 is available in Codex via both ChatGPT subscription and API key authentication, but API key users still lack access to cloud features (GitHub code review, Slack integration, cloud tasks), limiting the GPT-5.5 experience to local usage only.
- Cloud tasks are slower than local tools. Each task provisions a container, clones your repo, and installs dependencies before work begins, adding minutes of overhead compared to instant local execution.
How to get started
Choose your plan
Codex works with all ChatGPT plans, including Free. Plus ($20/mo) unlocks GPT-5.5 and cloud tasks. You can also use an API key for pay-per-token access with the CLI.
Pick your surface
Install the desktop app (macOS or Windows) for the richest experience, the CLI for terminal workflows, or the IDE extension for VS Code, Cursor, Windsurf, or JetBrains. All share the same configuration.
Connect your GitHub repository
Authorize Codex to access your GitHub repositories. For cloud tasks, connect via chatgpt.com/codex. For local work, the CLI and desktop app work directly with your local repo.
Create an AGENTS.md file
Add an AGENTS.md file to your repository root with setup commands, test commands, and project conventions. This tells Codex how your project works, similar to CLAUDE.md for Claude Code.
Describe your first task
Write a clear task description like "Add input validation to the signup form" or "Fix the failing test in src/utils/parser.test.ts." Codex works best with specific, concrete instructions.
(Optional) Set up mobile remote control
In the Codex desktop app sidebar, select "Set up Codex mobile," scan the QR code with your phone, and confirm the same ChatGPT account in the ChatGPT mobile app. You can then start and steer tasks from your phone.
Deep dive
Detailed guides with comparisons, tips, and visuals for each feature.
How Codex Cloud Works
The cloud sandbox model: how Codex spins up isolated environments, clones your repo, executes tasks in parallel, and delivers pull requests.
Codex CLI (Open Source)
The open source, Rust-built terminal agent with OS native sandboxing, MCP support, and multi-provider compatibility.
Surfaces and Remote Control
How Codex spans five interfaces (desktop, CLI, IDE, cloud, mobile) and how remote control lets you steer tasks from your phone.
Automations and Sandbox Model
Scheduled recurring agent tasks and OS native sandboxing that protect your machine at the operating system level.
AGENTS.md Configuration
Project level instructions that tell Codex your conventions, setup commands, and testing requirements.
Codex vs Claude Code vs Cursor vs GitHub Copilot
How OpenAI Codex compares to other AI coding tools across execution model, surfaces, autonomy, pricing, and extensibility.
Security and Isolation
How Codex provides security through OS native sandboxing, cloud isolation, approval policies, and review workflows.
Links
Apps
Documentation
Pricing
Similar Tools
Claude Code
anthropic
codingAnthropic's autonomous coding agent that lives in your terminal and understands your entire codebase.
ChatGPT
openai
chatbotThe most popular AI assistant in the world: text, images, video, voice, search, coding agents, and more in one place.
Cursor
anysphere
codingThe AI code editor and unified agent workspace built on VS Code that turns natural language into working software across your entire codebase.
Get notified about updates
We'll email you when this tool's pricing or features change.
Last updated: 2026-06-01