OpenAI released the Codex app for macOS, which lets developers run multiple AI coding agents simultaneously, each working on separate tasks in isolated copies of the same codebase.
To demonstrate its power, OpenAI had Codex build a complete racing game (eight maps, power-ups, multiple characters) from a single prompt. The agent consumed 7 million tokens, played the roles of designer, developer, and QA tester, and validated its work by actually playing the game.
The app also introduces Automations: scheduled tasks that run without supervision. OpenAI uses these internally for issue triage, CI failure summaries, and release briefs.
The company says more than a million developers have used Codex. Usage has doubled since GPT-5.2-Codex launched in mid-December. For a limited time, Codex is available to ChatGPT Free and Go users.
Greg Brockman, president and co-founder of OpenAI, tweeted, “i’ve been a die-hard terminal / emacs user for many years, but since using the codex app, going back to the terminal has felt like going back in time. feels like an agent-native interface for building.”
OpenAI built this app because IDEs were designed for human-to-code interaction, and that model is already obsolete for serious agent work. We are entering a new era of human/machine interfaces. So far, the chat window seems to be the winner.
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it. This work was created with the assistance of various generative AI models.