Codex

OpenAI's cloud-based coding agent that writes features, fixes bugs, and creates pull requests while running in isolated sandbox environments.

AI Tools

What is Codex?

Codex is OpenAI's cloud-based software engineering agent that handles complete coding tasks from start to finish.

It runs each task in its own sandbox environment preloaded with your repo, where it can write code, run tests, fix bugs, and propose pull requests for review.

Powered by codex-1 (a version of o3 trained on real-world coding tasks), it works through ChatGPT's sidebar or the new Codex desktop app. You assign it tasks, it works independently for 1-30 minutes, and you review the changes before merging.

Available to ChatGPT Pro, Business, and Enterprise users now, with Plus access rolling out soon. The desktop app for macOS lets you run multiple agents in parallel across different projects.

Good to Know

Runs tasks in isolated sandbox environments preloaded with your codebase

Powered by codex-1, a version of o3 fine-tuned on real-world development workflows

Can write features, run tests, fix bugs, and propose pull requests independently

Tasks typically take 1-30 minutes and you can monitor progress in real time

Available through ChatGPT sidebar and new macOS desktop app for managing multiple agents

How Vibe Coders Use Codex

Assigning it a feature request and reviewing the PR it generates an hour later

Asking it to explain a complex function in your codebase with context from surrounding files

Running multiple agents in parallel across different repos while you focus on architecture

Giving it a bug report and letting it iterate on fixes until tests pass

Frequently Asked Questions

AppWebsiteSaaSE-commDirectoryIdeaAI Business, In Days

Join 0 others building with AI