Summary created by Smart Answers AI
In summary:
- PCWorld reports that OpenAI has launched a native Windows desktop app for Codex, its AI coding tool that uses ChatGPT-powered agents to write code from natural-language prompts.
- Codex competes with similar tools like Anthropic’s Claude Code and Google’s Antigravity, offering adjustable autonomy levels from requiring approval to full automation.
- The tool supports complex project planning, connects to local directories and GitHub repositories, but high usage can quickly consume token allowances even with paid plans.
Crafting programs and apps with the help of AI agents has become quite the thing over the past few months, and Windows users just got a new “agentic” AI coding tool to play with.
Roughly a month after releasing its Codex desktop app for Mac, OpenAI has unleashed a native version for Windows, perfect for letting Windows users try their hand at coding programs and services with prompts.
Like Anthropic’s Claude Code app and Google’s Antigravity, OpenAI’s Codex lets you code with AI agents that can run in parallel, allowing you to direct multiple programming tasks simultaneously.
Codex was originally released as an open-source command-line tool last April, before the macOS-native desktop app landed in early February. Windows users had to wait a few more weeks for their own Codex version, but it’s finally here.
So, what’s Codex? Think of it as a workspace for AI agents who code for you based on your prompts, with the agents powered by specialized versions of ChatGPT.
Just tee up an existing coding project or an empty directory in Codex, and you can start coding with natural-language prompts, anything from “list the contents of this directory” to “make me an app that transcribes the recorded speech in audio files.”
Aside from writing specific lines of code, Codex and its AI agents can plan out their actions before actually performing them, thinking through complex coding projects and offering detailed roadmaps that you can tweak and comment on before you give the go-ahead.
Before setting your Codex AI agents loose, you can choose their level of autonomy. On one end of the spectrum, you can make them come back for approval before every command; on the other end, you can take your hands off the wheel and let them code, although doing so can be both unpredictable and expensive.
Aside from locally-hosted directories, you can connect Codex to remote GitHub repositories, perfect for letting your agents check out “branches” of GitHub projects before doing any tweaking. Codex also lets you set up worktrees that keep the work of your AI agents in a sandbox before being deployed to a live production environment.
A left-hand column in the Codex app lets you keep tabs on multiple Codex chats at once, meaning you can have teams of AI agents working on separate projects all at the same time. Notifications will pop up for each chat when a pending agent action needs your approval.
Competing with Codex is Claude, which also has a command-line (CLI) version as well as a desktop component within the main Claude app. A newer player in the agentic AI coding field is Google Antigravity, which lets you code with agents powered by Google’s Gemini models as well as Claude and OpenAI’s open-weight GPT-OSS models.
Like Claude Code and Antigravity, OpenAI’s Codex is free to use and works with free ChatGPT accounts, but you’ll have to contend with strict usage quotas. Even with paid ChatGPT Plus ($20 a month) and Pro ($200 a month) plans, your Codex allowance may quickly run out, given the rate at which AI coding agents burn through tokens.