From December, ChatGPT's Plus, Pro and Business subscriptions will share a single token budget across ChatGPT and Codex. It is the step with which OpenAI merges the Chat and Work modes, bringing Work's agentic features into the main interface. For people who use the assistant every day, the way usage is measured changes: no longer by the number of requests, but by the amount of information the model processes.

Psst!Add AlexTech.ai asPreferred Source on Google

How the new budget works

Tokens measure the information a model processes and generates. The more complex the model and the more involved the task, the faster the budget runs down. Users can check their remaining balance and reset time in Settings. Once they hit the limit, they can wait for a reset, upgrade to a higher plan or buy extra credits.

official image of ChatGPT

Image: chatgpt.com — https://chatgpt.com/overview/

Chats with Instant models do not count toward the budget. On paper ChatGPT therefore remains a flat-rate service, but heavy users of the cheaper plans risk getting lower-quality results. OpenAI recently revised the limits of its Pro and Business plans, doubling the quotas as already reported by AlexTech.

Why it changes now

The move follows the convergence of the two working modes. Today users choose between two operating modes and eleven model settings, a complexity even experienced users struggle to follow. The new system simplifies the choice but shifts the constraint from the number of messages to the complexity of the work performed.

The change comes as OpenAI pushes agentic work, already visible in the five-hour limits for Codex and ChatGPT Work and the new data agents. When Codex and Work share the same budget, a long programming task can reduce what is left for ordinary conversations.

What changes for subscribers

Subscribers to Plus, Pro or Business will need to keep an eye on their token balance, especially if they often use advanced models or agents. Those who use ChatGPT mainly for simple chats with Instant models should notice no difference. OpenAI has not yet explained how the budget will translate into precise figures for each plan, or how much extra credits will cost.