The all-you-can-prompt buffet is getting a tab. Starting in December, OpenAI will put ChatGPT and its Codex coding agent on one shared allowance, so your late-night refactor and your afternoon chat about quarterly numbers now eat from the same plate.

One Budget, Two Very Hungry Apps

According to an updated OpenAI support page reported by The Decoder, Plus, Pro and Business subscribers will share a single token budget across ChatGPT and Codex. Tokens measure how much information a model processes and generates, so heavier models and more complex tasks burn through the allowance faster. Users will be able to check their remaining balance and reset times in Settings, and when they run dry they can wait for a reset, upgrade, or buy extra credits.

There's an important carve-out: chats with OpenAI's Instant models won't count against the budget. The change comes as OpenAI folds Work's agentic features into the main ChatGPT interface, and it mirrors a move Anthropic already made when it combined Chat and Cowork in Claude.

Flat-Rate AI Was Always a Bit of a Fiction

On paper, ChatGPT stays flat-rate, since everyday Instant chats remain uncounted. In practice, heavy users on cheaper plans will likely get nudged toward lighter models once the good stuff runs out. Agentic work like long coding sessions is expensive to serve, and pooling it with chat makes that cost visible to the people generating it.

Here's the part teams should notice. If the same people use ChatGPT for writing and Codex for coding, one ambitious agent session can drain the allowance they were counting on for everything else that week. A token budget is a cost model, and any business rolling AI out across departments now has to treat usage the way it treats cloud spend: monitored, forecast, and occasionally the subject of an awkward conversation about who kicked off the giant refactor on a Friday afternoon.

It also changes how you pick tools. When every task draws from the same pool, choosing the lighter model for routine work stops being a nerdy optimization and becomes basic budgeting.

Unlimited AI isn't disappearing so much as finally handing over the receipt.

If a shared token budget has you rethinking how your team actually uses AI day to day, we can help you build a rollout plan that fits both the work and the bill.

Source: The Decoder