Codex Dev Stack
AdvancedOpenAI Codex on your desktop, in your terminal, or in the cloud, steerable from your phone.
Published 27 September 2026
About Codex Dev Stack
Codex is OpenAI's coding agent: it takes a task in plain language, works through the repository, writes and runs code, iterates on failures, and hands back a diff or a pull request for review. It's built around OpenAI's models, which ChatGPT plans cover, so OpenAI sits in this stack's core; its configuration also supports other providers, such as OpenRouter or local models, billed through their own keys.
The same agent ships on several surfaces. The ChatGPT desktop app on macOS, Windows, and Linux runs several Codex projects in parallel; the CLI works from any terminal; an IDE extension puts it beside the code in the editor; and Codex cloud runs tasks in OpenAI's own sandboxes. All of them are included in ChatGPT plans, from limited use on the free tier up.
Work keeps going away from the desk in two ways. Codex cloud runs tasks in parallel on OpenAI's infrastructure, so your machine can be off, and delivers each result as a branch or pull request; it connects to GitHub, with GitLab support in beta. The ChatGPT app on iOS and Android also steers Codex on your own Mac or Windows computer running the desktop app, including work inside SSH hosts it's connected to, as long as that computer stays awake and online. Some developers also run the CLI on an always-on VPS or home server of their own.
Code lives on GitHub by default, or on GitLab when source control has to be self-hosted. A faster terminal emulator for the CLI, Docker for a reproducible environment, and a CI/CD pipeline are common additions; none of them change how Codex works.
Key Features
- ✓One agent across the ChatGPT desktop app, the CLI, an IDE extension, and Codex cloud
- ✓Parallel tasks in OpenAI's cloud sandboxes, each delivered as a branch or pull request
- ✓Steering from the ChatGPT iOS and Android apps for work running on your own Mac or Windows computer
- ✓Remote SSH in the desktop app for projects that live on a devbox or server
- ✓Configurable cloud environments with setup scripts and custom dependencies
- ✓Included in every ChatGPT plan, with limited use on Free and Go
When to Use Codex Dev Stack
- →Queuing several well-specified tasks in Codex cloud and reviewing the pull requests later
- →Running Codex in the terminal alongside an existing editor setup
- →Approving an action or checking test output from a phone while Codex works on the desktop
- →Working on a remote devbox through the desktop app's SSH connection
- →Trying an agent-first workflow on a ChatGPT plan you already pay for
Pros
- Parallel cloud tasks run without using the local machine
- The same subscription covers ChatGPT and every Codex surface
- A free tier and an $8 Go plan make it cheap to try
- Pull-request delivery keeps human review on everything merged
Cons
- Plan usage covers OpenAI's models only; other providers need their own keys and configuration
- Cloud tasks run code through OpenAI's infrastructure, which some teams can't allow
- Phone access to a local session needs the desktop app on macOS or Windows, not a Linux host
- Results depend heavily on how precisely the task is specified
Version Control Options for Codex Dev Stack
Codex Dev Stack with GitHub
GitHub is the default here: the largest pull-request ecosystem, and where Codex's own PR-delivery workflow is most commonly wired up.
Codex Dev Stack with GitLab
GitLab bundles source control, merge requests, and CI/CD in one self-hostable platform, the pick for teams that need code and pipeline data to stay on their own infrastructure.
Codex Dev Stack Add-ons
Each addition below extends this stack with a capability the base stack works fine without. None are required: include the ones your product actually needs when building this stack, and skip the rest.
Server Add-ons
Add a server when you want the agent to keep running somewhere other than your own laptop, reachable at any time and not tied to your machine staying on. Some agents also offer their own managed cloud sessions as an alternative to self-hosting; check the stack's own description for details.
Codex Dev Stack with Hetzner
Hetzner is the default here: a 4 GB server, the Codex CLI's documented minimum, is about €5.49/mo, with 8 GB plans for the recommended size. A plain VPS, the cheapest way to keep a Codex CLI session running.
Codex Dev Stack with DigitalOcean
DigitalOcean offers a 1-Click Codex CLI Droplet with the agent preinstalled; the 4 GB size is $24/mo and 8 GB $48/mo. Its Managed Agents, in public preview, also run the Codex CLI in isolated microVMs billed per second, an alternative to a whole server.
Codex Dev Stack with Hostinger
Hostinger's KVM 1 has 4 GB of RAM at $6.49/mo introductory, renewing at $11.99/mo; KVM 2 doubles that to the recommended 8 GB. A budget pick for one Codex host, on a plain Linux image.
Codex Dev Stack with Raspberry Pi
A Raspberry Pi 5 with 4–16 GB of RAM is a one-time hardware cost instead of a monthly bill. The Codex CLI ships ARM64 Linux builds; an 8 GB board meets the recommended memory with room for tests.
These are highlighted picks. To see all the tools, check the Hosting & Cloud category.
Remote Access Add-ons
Add remote access when you want to reach an agent running on another machine — a VPS or a home server — without exposing it to anyone but you.
Codex Dev Stack with Tailscale
Add Tailscale alongside a server to reach the Codex CLI privately over SSH from your laptop or phone without exposing the machine to the public internet. The personal tier is free.
Session Persistence Add-ons
Add session persistence when the agent runs on a server or over SSH: a terminal multiplexer keeps the session working after the connection drops or the laptop closes, and you reattach from any machine to pick up where it left off.
Codex Dev Stack with tmux
Add tmux when Codex CLI runs on a server or over SSH: closing the terminal otherwise ends the run, while a tmux session keeps the agent working after you disconnect and lets you reattach from any machine. It's already installed or one package away on most Linux servers.
Codex Dev Stack with Zellij
Add Zellij when you want the same keep-running Codex sessions with an easier start: its status bar shows every shortcut, closed sessions come back with their layout after a reboot, and a built-in web client opens a running session from a browser.
Terminal Add-ons
Add a terminal when you want a faster, more configurable place to run the agent than your OS default — most agent CLIs live here all day.
Codex Dev Stack with Ghostty
A GPU-accelerated terminal with fast startup and native platform integration on macOS and Linux, a solid default for running the Codex CLI all day.
Codex Dev Stack with Alacritty
A minimal, GPU-accelerated terminal focused purely on speed with almost no built-in features beyond rendering; pick this if a separate multiplexer or window manager already handles everything else.
Codex Dev Stack with WezTerm
A GPU-accelerated terminal with built-in multiplexing, tabs, and a Lua-based config, useful for one tool that replaces both the terminal and tmux.
Codex Dev Stack with Kitty
A GPU-accelerated terminal with its own graphics protocol and scripting API, popular with developers who want deep customization through its kitten plugin system.
Codex Dev Stack with iTerm2
A free, open-source macOS terminal with more than a decade of features: split panes, profiles, triggers, and tmux integration. The familiar pick for Mac users who want the Codex CLI in the terminal they already know.
Code Review Add-ons
Add code review when you want an AI reading every pull request before it merges: it comments on the diff so bugs, security issues, and inconsistencies surface before a human has to catch them.
Codex Dev Stack with CodeRabbit
Add CodeRabbit when you want an AI reviewer on every pull request: it installs as a GitHub or GitLab app, summarizes the diff, and leaves line-level comments before anyone merges, with a free tier for open-source repos.
Codex Dev Stack with Greptile
Add Greptile when you want review that reads the whole repository rather than just the diff: it indexes the codebase so its GitHub and GitLab pull request comments carry wider context, and it can be self-hosted.
Codex Dev Stack with Cursor Bugbot
Add Cursor Bugbot when a bug-focused reviewer is enough: it reviews GitHub or GitLab pull requests whatever wrote the code, though it bills through a Cursor account and plan rather than its own subscription.
CI/CD Add-ons
Add CI/CD when you want a dedicated pipeline for running tests, linting, or multi-stage builds before a deploy goes out. Many hosting platforms already redeploy automatically on every push on their own — a CI/CD tool adds the most value on top of that by gating the deploy on a passing test suite, and matters even more when the hosting choice does not auto-deploy at all, such as a self-hosted server.
Codex Dev Stack with GitHub Actions
Add GitHub Actions when the repo lives on GitHub and you want tests and deploys running automatically on every pull request Codex opens.
Codex Dev Stack with GitLab CI/CD
Add GitLab CI/CD when the project is on GitLab and needs the same automated pipeline, defined in a .gitlab-ci.yml file that can run entirely on your own infrastructure.
These are highlighted picks. To see all the tools, check the CI/CD Pipelines category.
Containerization Add-ons
Add containerization when you want the app packaged the same way across local development, staging, and production, or need to deploy somewhere that isn't a managed serverless platform.
Codex Dev Stack with Docker
Add Docker when a task needs a reproducible execution environment beyond what the default cloud sandbox provides, or when the same environment should run identically locally and in CI.
These are highlighted picks. To see all the tools, check the Containerization category.
Frequently Asked Questions about Codex Dev Stack
Can Codex keep working when my computer is off?
Yes, through Codex cloud. Tasks run in OpenAI's sandboxes, several at once, and come back as branches or pull requests, so your own machine can be off; it needs the repository on GitHub, or on GitLab while that support is in beta. The ChatGPT mobile app is the other route: it steers Codex running in the desktop app on your Mac or Windows computer, but that computer has to stay awake, online, and signed in. A Codex CLI on an always-on server of your own is the option for developers who want the environment under their control.
How does Codex compare to Claude Code?
Both are standalone coding agents with a desktop app, a CLI, IDE extensions, and a cloud mode, so the practical difference is the model family and the plan: Codex is built around OpenAI's models on a ChatGPT plan, and its config can add other providers; Claude Code runs Claude on a Claude plan. Codex has a free tier and an $8 Go plan; Claude Code starts at $20. Codex's phone access needs the desktop app on macOS or Windows as host, while Claude Code's Remote Control works with a terminal session on any system, Linux included.
Which Codex surface should I use?
They're the same agent, so it comes down to how you like to work. The desktop app suits running several projects in parallel and is the host for phone access. The CLI suits developers who live in the terminal or want Codex on a server. The IDE extension keeps it beside the code in VS Code and its forks. Codex cloud suits tasks you want to hand off and review later.
Which plan do I need for heavy use?
Free and Go are for trying Codex on quick tasks. Plus covers regular daily use; Pro raises Codex limits five or twenty times, for long sessions and many parallel cloud tasks. Business adds workspace controls for teams. API-key billing charges per token instead of plan limits, which suits automation. The Pricing section has the current figures.
What goes wrong with many parallel cloud tasks?
Each task works on its own branch in its own sandbox, so they don't collide while running, but two tasks that touch the same files produce pull requests that conflict with each other at merge time. The other limit is review: several tasks finishing at once can produce more changes than one person can read carefully, so parallelism pays off most on independent, well-specified work.
Scores
Popularity4/5
One of the two most widely used standalone coding agents, helped by reaching every ChatGPT subscriber.
Learning Curve2/5
A plain-language task in the desktop app or cloud is an easy start; getting consistent results takes practice writing precise tasks and reviewing diffs.
Flexibility4/5
Desktop, CLI, IDE, cloud, and phone access, plus configurable cloud environments, within OpenAI's own model family.
Performance4/5
Parallel cloud tasks move work off the local machine; each individual task takes minutes rather than seconds.
Portability4/5
Works on plain files and git; cloud environment setup and task habits are Codex-specific, but little code is.
Tools in the Codex Dev Stack Stack
Codex Dev Stack Pricing
Codex comes with ChatGPT plans: Free and Go ($8/month) for limited use, Plus at $20/month for regular use, and Pro at $100 or $200/month for 5x or 20x Plus's limits. Business is $25 per seat monthly, and API-key billing charges per token instead. GitHub or GitLab are free for individual developers. The optional server runs about €5.49–24/month for 4 GB, and Tailscale's personal tier is free.
Free and Go ($8/mo) are limited; Plus $20/mo; Pro $100/mo (5x) or $200/mo (20x, new sign-ups paused); Business $25/seat; or API-key billing.
Free tier covers individual devs; GitHub Team is $4/mo, GitLab Premium is $29/mo for more seats.
Docker Desktop is free for personal use; Docker Pro adds team features from $5/mo.
Tailscale's personal tier is free for up to 100 devices.
A 4 GB VPS (8 GB recommended): Hetzner €5.49, Hostinger $6.49 intro ($11.99 renewal), DigitalOcean $24; or a Raspberry Pi as a one-time cost.