Eshu Marneedi

OpenAI, on X:

Introducing ChatGPT Work, a new agent in ChatGPT powered by Codex and GPT-5.6. It can take action across your apps and files, stay with a project for hours if needed, and turn a goal into finished work. It’s a whole new way to get work done.

ChatGPT Work is powered by GPT-5.6. GPT-5.6 makes ChatGPT state of the art at reasoning through complex tasks and creating materials that match your templates, reference files, and preferred style. Just describe the outcome you want, without having to spell out every step to get there.

On web and mobile, ChatGPT Work is rolling out today for Pro, Enterprise, and Edu plans. It will roll out to Plus and Business plans over the next few days. In the ChatGPT desktop app, Chat, Work, and Codex are available on every plan, including Free, and is available globally to download on Windows and Mac.

Codex app users can update their app as usual — it will become the new ChatGPT desktop app.

I fear OpenAI’s employees have lost the plot. The old, native ChatGPT app for macOS still exists and is renamed “ChatGPT Classic” after updating to the latest version. The Codex app, after updating, is renamed “ChatGPT.” The unfortunate side effect of this is that if you update the Codex app before the ChatGPT Classic app, the two names will conflict, and macOS will refuse to open the Codex app. This happened to me; I had to reinstall the Codex app from the OpenAI website to fix the issue.

Now I have two ChatGPT apps on my Mac: ChatGPT Classic and ChatGPT, née Codex. The ChatGPT Classic app has languished for years now, but when it was announced, it truly was incredible. It’s beautifully crafted with care, down to the little details. There’s even a mode to let ChatGPT peek into your apps like BBEdit or TextEdit — this is how we used to code with artificial intelligence before Codex and Claude Code! The Codex app is made with Electron, but I’ve always regarded it as one of the best Electron apps on the Mac. It’s not perfect, but superior to Claude in every dimension.

The new update ruins months of good favor with the Codex app. It quite literally benefits nobody. There are now two primary modes in the ChatGPT app: ChatGPT Codex and ChatGPT Work. Codex is the same; Work is just a simpler version of Codex, similar to Claude Cowork. I don’t even know why this mode exists — Codex has always been simple enough for non-technical people to use anyway. And I don’t understand how ChatGPT Work can function on the web or in the iOS app, since it works with files on-device, not in the cloud. (I don’t have a ChatGPT Pro subscription, so I’ll have to wait a few days to try it out.)

The actual ChatGPT interface — what the app is named for! — is hidden in a mini-window accessed from the sidebar. This is the most absurd, counterintuitive user interface OpenAI has ever conceived. If the app is literally named “ChatGPT,” how is ChatGPT not the main interface? Does OpenAI see ChatGPT Work as the replacement for ChatGPT, and if so, why doesn’t it make that clear? And to make matters worse, there’s no consistency in the models across interfaces. Work and Codex have GPT-5.6 Sol, Luna, and Terra — the three sizes of GPT-5.6 — while ChatGPT only exposes the Instant, Medium, and High reasoning levels. Medium and High are GPT-5.6 Sol, though that is not indicated in the interface.

The ChatGPT app doesn’t open ChatGPT anymore, and the models once again make no sense. Here’s a handy chart of the models and their reasoning options across ChatGPT modalities:

I haven’t heard from a single person happy with Thursday’s changes. Codex users are upset because their app has been renamed; ChatGPT users on the Mac are upset because the new ChatGPT app really isn’t ChatGPT at all. The only people seemingly pleased with this “work” are OpenAI employees themselves, who seemingly don’t understand anything about marketing or communications. But I’m supposed to believe this company is worth a trillion dollars.