If you’ve ever opened the ChatGPT desktop app and wondered which tab your project lives in, OpenAI has noticed too. That’s a real design problem for a product used by hundreds of millions of people, and the fix is overdue.
OpenAI is merging the Chat and Work tabs, according to TechCrunch, which cited an email sent to users. The app currently splits into ChatGPT and Codex, and ChatGPT itself splits again into Chat and Work. That’s three places to keep track of, and it’s easy to lose a thread. The merged experience will go to a select group of users for feedback first, then roll out more widely.
The pitch in the email is straightforward. You start by exploring an idea, then ChatGPT carries it through, all in one place. You bring your files and plugins, ask it to build a website, make a presentation, or do tasks on the web, and you review and steer the work as it goes. Codex stays a separate experience inside the desktop app.
The more interesting detail is buried lower. ChatGPT and Codex will now draw from the same token allowance pool. OpenAI says most users won’t notice. But depending on your usage and which model you pick, some people will hit limits sooner and wait for resets more often. If you lean on Codex for long coding sessions and also use ChatGPT for agent-style tasks, that’s you. Heavy users should watch their usage closely after the change.
This also isn’t happening in a vacuum. Anthropic merged Claude’s chat and Cowork functions into one screen in September, so OpenAI is following a rival here, not leading. The direction is clear across the industry. Chat is no longer the product. The product is an assistant that takes a vague request, uses your files and tools, and finishes the job. Separate tabs for talking and doing undercut that story, because the user shouldn’t have to decide in advance which mode a task belongs in.
Does it matter? For casual users, mildly. A cleaner interface is nice, nothing more. For developers, founders, and teams who use ChatGPT for real work, it matters more, because it signals where OpenAI wants the app to go: agent first, chat second. The shared allowance is the part to watch. A tidier interface is easy to like. A tighter usage budget is less so, especially if you’re paying for a plan and expect Codex capacity to stay separate.
Keeping Codex distinct is a sensible call. Coding workflows have different needs than slide decks and web tasks. Still, sharing one meter between them suggests OpenAI is treating compute as a single resource to manage, and users will feel that before they feel any benefit from the new layout.
