OpenAI plans to combine ChatGPT and Codex usage into a single token allowance for its Plus, Pro, and Business subscriptions. According to a company support page, this change is announced for December and is part of the transformation of ChatGPT into a work platform equipped with agentic capabilities.
- OpenAI prévoit de fusionner les quotas de ChatGPT et Codex pour les abonnements Plus, Pro et Business.
- Le changement est annoncé pour décembre et s’inscrit dans la transformation de ChatGPT en plateforme de travail.
- Les modèles Instant ne seront pas décomptés de ce quota commun.
- L’application ChatGPT d’origine disparaîtra le 30 janvier 2027 au profit d’une version intégrant davantage les fonctions agentiques.
The goal is to bring the Chat and Work environments closer together and simplify the management of usage across the different features of the OpenAI ecosystem. Until now, users had to juggle multiple environments and separate consumption rules. The token, the unit of text processed by the models, serves as the basis for measuring this consumption. The consolidation is intended to pool the resources used in ChatGPT and Codex.
A unified consumption between conversation and agents
This convergence reflects a fundamental shift in the company’s strategy. ChatGPT is no longer limited to Q&A: the platform is gradually incorporating functions capable of executing tasks, interacting with tools, and handling more complex work sequences. In this context, maintaining a strict separation between conversational and agentic uses is, according to the source, becoming less relevant. Work features are intended to integrate more fully into the main ChatGPT experience.
The shared quota is meant to accompany this convergence. Operations performed in the different environments would thus draw from the same consumption allowance, according to the terms set by the publisher.
Increased limits, but consumption to monitor
OpenAI has also reportedly doubled the usage limits for some high-end plans, which should provide more headroom for users who regularly call on the models for complex tasks. Token consumption, however, remains a central issue. A simple query does not use the same resources as an agentic operation involving multiple steps, calls to tools, or processing large volumes of data.
Users will be able to check their balance and the renewal date of their allowance in the settings. If the quota is exhausted, three options would be offered: wait for its reset, upgrade to a higher plan, or purchase additional credits. An exception is nevertheless planned: exchanges using the Instant models would not be deducted from this common allowance.
The unification would not only concern quotas. OpenAI is also preparing an evolution of its application, with the ambition of bringing the different experiences together within a more integrated environment. The original ChatGPT app is set to disappear as of January 30, 2027, in favor of the new version. Users will then have to adopt this renewed interface, which is intended to integrate conversational and agentic features more tightly.
