I tested the new OpenAI Codex features on a real Python codebase, and it’s the strongest Claude Code rival yet

OpenAI expanded Codex with an in-app browser, Mac computer-use support, PR review, SSH access, and plugins.

In tests on the open-source HTTPie project, Codex used the browser to read a GitHub issue and fix the bug in minutes.

It also handled some Mac GUI navigation, though with limits, and reviewed a pull request by checking docs and running tests.

The update makes Codex more useful across real development workflows and strengthens its position against other AI coding assistants.