What is Codex?

OpenAI's coding agent. The same idea as Claude Code, wearing an OpenAI badge, and probably already bundled in the ChatGPT plan you pay for every month.

There is a good chance you are already paying for Codex and have never opened it. It ships inside ChatGPT Plus, the $20-a-month plan most CEOs already have, and inside Pro, Business, and Enterprise. It is the most capable tool sitting unused in your subscription right now.

What it is

Codex is OpenAI's coding agent. Their own one-line pitch for it is "one agent for everywhere you code," and the "everywhere" is the point. The same Codex shows up in five places: in your terminal as the open-source Codex CLI, as an extension inside code editors like VS Code and JetBrains, on the web as a cloud agent you hand a task and walk away from, inside ChatGPT itself, and wired into GitHub to open and review pull requests.

In every one of those places it does the same job. It reads a codebase, edits files, runs commands and tests, reviews code for bugs, and can grind on a single task for a long stretch without you babysitting it.

OpenAI shipped the Codex CLI and the cloud version in 2025 and has swapped the engine underneath it roughly every month since. It started on a model called codex-1, tuned for software work, and now runs on OpenAI's GPT-5 family of coding models. The CLI is free and open source. If you want to see it, the install is one line: npm i -g @openai/codex.

The cleanest way to hold it: Codex is OpenAI's answer to Anthropic's Claude Code. Same category, opposite vendor.

Why CEOs care

Three reasons, in order of how much they should change your week.

First, you likely already own it. Codex is bundled into ChatGPT plans, so there is no new vendor, no new bill, no procurement. The cost of trying it is the ten minutes it takes to find it.

Second, your engineers are choosing between Codex and Claude Code right now, and the choice shapes how fast your product moves. Knowing what both are lets you ask a sharper question than "which is better": "which one fits how we work."

Third, and this is the load-bearing one: the skills you build are portable, so you are not locked in. Codex, Claude Code, and most of the field now read the same open Agent Skills format. A workflow you write once runs on either agent. That means you can pick a tool this quarter without betting the company on it, and switch when the lead changes hands. And it changes hands constantly. The two are roughly at parity and trade the top spot every few weeks.

Where you'll see it

• Inside the ChatGPT subscription you already pay for, under the Codex tab.
• In your engineers' terminals, going head to head with Claude Code.
• In the "Codex vs Claude Code" thread your team has probably already started.
• In any conversation about agents that run unattended in the cloud and open pull requests while you sleep.

Codex or Claude Code?

The honest operator answer: it matters less than the internet wants it to. They are close enough that picking the "wrong" one costs you almost nothing, and switching later is cheap because your skills travel with you.

I run Claude Code and OpenCLAW, and the reason is the harness ecosystem around it, not a benchmark score. Your reason might be that Codex is already in your ChatGPT plan and your team lives in OpenAI's world. Both are fine. The mistake is tab-shopping: trialing five tools, mastering none. Pick one, get genuinely fluent, keep your skills portable, and switch the day it stops being the best. If you want my full side-by-side, Codex vs Claude Code is the head-to-head.

What to do next

Open the ChatGPT plan you already pay for, find the Codex tab, and give it one real task you would otherwise hand to an engineer. If you would rather start with the version this site is built around, read What is a coding agent? for the category, then What is OpenCLAW? for the harness most operators run. Either way, watch one work end to end this week. I would love to hear what it shipped for you.