What is OpenAI Codex?

OpenAI Codex is OpenAI's AI coding agent — a tool that reads your codebase, writes and edits code, runs commands, and works through programming tasks you describe in plain English.

The name has some history. The original Codex was a 2021 model that powered the first GitHub Copilot; OpenAI later retired it, then brought the name back for something more ambitious: an agent that doesn't just suggest lines but actually does the work.

As of July 2026, Codex shows up in several places, all sharing the same underlying model and account:

• Codex CLI — an open-source agent that runs in your terminal, editing files and running commands in a sandbox on your machine.
• IDE extension — the same agent inside editors like VS Code.
• Codex cloud — tasks you hand off through ChatGPT that run on OpenAI's servers and can come back as ready-to-review pull requests.

The cloud piece is the interesting part: you can queue up several tasks in parallel — fix this bug, write these tests, update that dependency — and review the results later, like delegating to a very fast junior developer who never sleeps.

Its main rivals are Anthropic's Claude Code and agent-mode editors like Cursor. The honest take: all three are good, and teams often pick based on which model family they already trust. If you're choosing, see Cursor vs Copilot vs Claude Code.