OpenAI merges ChatGPT Work into Chat

For a lot of ChatGPT users, the awkward part of getting real work done has never been the model itself. It has been deciding which mode to live in, and then discovering that the push toward more advanced tasks quietly eats into a separate budget. OpenAI now says it is preparing to change both of those things at once, by pulling its Work features into ordinary Chat and making ChatGPT and Codex draw from a single pool of usage.

The company confirmed the plan in an email to ChatGPT subscribers, which BleepingComputer reviewed. The message describes a test that will bring the Chat and Work tabs together for a limited set of users, alongside a rework of how Pro allowances are counted.

One tab instead of two

According to the email, OpenAI will test a combined experience for the Chat and Work tabs over the coming weeks. The rollout is not a general release from day one: the company says it will begin with a small group of users before widening access.

The point of merging the two interfaces is continuity. In the current setup, a user moves between conversational ChatGPT and the more task-oriented Work environment depending on what they are trying to accomplish. Under the plan described in the email, a person selected for the test can travel from an everyday conversation into heavier work without changing modes.

OpenAI framed the change as keeping a single thread of work intact from start to finish:

"This will enable you to move from exploring an idea to having ChatGPT carry it through, all in one place,"

— OpenAI, in an email to ChatGPT subscribers

The email also sketches what that combined environment is meant to handle. Users would be able to bring in their files and plugins, then ask ChatGPT to build a website, put together a presentation, or carry out tasks on the web, while the user reviews and guides the work as it happens.

That description places the change squarely in the territory of agentic assistance: the assistant is not only answering, it is executing multi-step jobs with the user supervising. Consolidating that into the main Chat surface removes a mode switch that, for many subscribers, has been a source of confusion about which features live where.

Codex keeps its own front door

One boundary is staying in place. OpenAI said Codex will remain a separate experience within the ChatGPT desktop app, even as the underlying usage accounting is unified. The distinction matters for anyone who assumed the announcement meant Codex was being absorbed into the Chat interface.

So the change is split in two directions. On the surface, Chat and Work converge. Underneath, ChatGPT and Codex converge on the same allowance. But the Codex product itself keeps a distinct home in the desktop client.

For subscribers, that means the practical effect of the announcement will depend on which part they use. Someone who mostly works in the browser will notice the tab merge. Someone who leans on Codex in the desktop app will feel the shared-limit change more directly.

The shared allowance, explained

The second half of OpenAI's plan concerns Pro subscriptions. The company said it is preparing to unify usage across ChatGPT and Codex so that both services draw from the same pool.

"Your use of ChatGPT and Codex will draw from the same allowance,"

— OpenAI, in an email to ChatGPT subscribers

Not every interaction will register against that pool. OpenAI said Unlimited Instant is included in the subscription and will not count toward the shared usage limit. More advanced models and complex tasks, by contrast, will consume more of the allowance depending on the work involved.

That structure mirrors how tiered AI usage has generally been metered: light, fast interactions are treated as effectively free, while heavier reasoning and agentic jobs carry a higher cost. The difference here is that the meter is now shared between two products that previously kept their own books.

Users will be able to see where they stand. OpenAI said the combined remaining usage and reset times will be visible under Settings > Usage. If the allowance runs out, the email lists the available paths: wait for it to reset, upgrade the subscription, or purchase credits for eligible features.

What Pro users are getting right now

Separately from the future test, OpenAI said it has already doubled Work and Codex allowances for its Pro 100, Pro 200, and Pro 500 subscriptions, effective October 29. That increase applies to the tiers named in the email and is described as already in effect rather than part of the upcoming experiment.

The timing is worth separating out. The allowance doubling is a present change. The combined Chat and Work experience and the unified ChatGPT-Codex usage pool are still ahead, with the former explicitly beginning as a small-group test over the coming weeks.

• Pro 100, Pro 200, and Pro 500 subscriptions have had Work and Codex allowances doubled, effective October 29
• The combined Chat and Work test starts with a small group of users before expanding
• Unlimited Instant is included in the subscription and does not count toward shared usage
• Combined usage and reset times appear under Settings > Usage

Why a shared meter is a real change

Until now, a subscriber could reasonably think of ChatGPT usage and Codex usage as two separate budgets. If one ran dry, the other was still there. A unified allowance removes that cushion. Heavy use of one product now has a direct effect on the other.

For teams that have standardized on Pro tiers, that changes how they plan. A developer who spends a day pushing complex Codex tasks is drawing from the same account balance as the colleague writing documents in Chat. The Settings > Usage readout becomes a shared dashboard rather than a per-product gauge.

The credit path matters too. The email states that running out leaves three options: waiting for a reset, upgrading, or buying credits for eligible features. That is a straightforward set of levers, but it puts the onus on the user to monitor the shared pool before a deadline or a project crunch arrives.

The trial runs small on purpose

OpenAI's decision to start the combined experience with a small group fits the pattern of how it has handled interface changes in the past: limited exposure first, then broader access. The email does not give a date for expansion beyond the coming weeks, and it does not specify how users will be selected for the initial group.

What the email does make clear is the intended end state. The combined interface is described as letting a user bring files and plugins into a single place and direct ChatGPT through website builds, presentations, and web tasks, reviewing as it goes. The two-tab split, in that framing, is an artifact the company is working to remove.

Codex's separate status in the desktop app is the exception that keeps the picture from being a full merge. The usage pool is shared; the product surfaces are not entirely.

What subscribers should watch

The most immediate thing for Pro subscribers to check is the Settings > Usage screen, which will now reflect combined remaining allowance and reset timing. Anyone on the Pro 100, Pro 200, or Pro 500 tiers should also see the doubled Work and Codex allowances that took effect on October 29.

For users who rely on both ChatGPT and Codex, the shared pool is the change that deserves attention before the test reaches them. The practical question is no longer "how much ChatGPT do I have left" but "how much do I have left across everything."

Those selected for the combined Chat and Work test will be the first to see whether folding Work into the main interface actually delivers the continuity OpenAI describes. Everyone else will be watching the allowance numbers.

Why it matters

The direction here is toward fewer modes and one meter, which is convenient on paper but shifts risk onto the user. A shared allowance could mean that an intensive Codex session leaves less room for everyday Chat work, and the only warning will be a settings page. Organizations that budget AI access by seat may find they need to think about usage patterns across tools rather than per product. If the combined experience works as described, it could also lower the friction that has kept some users from attempting more complex tasks in ChatGPT at all. Either way, the test is small for now, but the accounting change behind it applies to the subscriptions OpenAI has already named.

Reporting based on original coverage from BleepingComputer.