ChatGPT and Codex now share the same usage quota
OpenAI combines ChatGPT and Codex quotas starting December for Plus, Pro, and Business subscribers. The change simplifies usage but worries developers who fear losing capacity.
Starting in December, OpenAI will unify the usage quotas of ChatGPT and Codex for its Plus, Pro, and Business subscribers. Practically, ChatGPT conversations and Codex tasks will draw from a common pool, as the company seeks to bring its conversational tools and agents closer together. This change mainly affects developers and heavy users, as a resource-intensive coding session can now reduce chatbot usage, and vice versa.
Une expérience plus fluide, mais moins cloisonnée
The logic is simple on paper: OpenAI wants to avoid making users juggle between multiple modes and products. ChatGPT Work, launched in July, was already designed to delegate complex tasks such as research, data analysis, or producing documents, presentations, and spreadsheets. Alongside this, the company has brought ChatGPT and Codex closer in a unified desktop application, with dedicated interfaces for conversations, agent tasks, and coding.
This usage merger goes beyond a simple visual grouping. The separate ChatGPT and Codex quotas disappear to become a shared ceiling, the exact size of which OpenAI has not yet specified. Instant mode conversations will remain excluded from this count, but heavier tasks will depend on the chosen model and the complexity of the work requested. For subscribers, the change could be very tangible. For Excel spreadsheets, less so—unless the spreadsheet suddenly demands a lawyer.
Les développeurs regardent surtout le compteur
On OpenAI’s developer forum, several subscribers have requested the maintenance of separate limits, or at least more transparency on consumption calculations. One user, Smalikou, writes: “ChatGPT and Codex serve different purposes, and their usage limits should remain separate.” Another, PZA, calls for a real-time dashboard showing balance, reset time, and consumption by product, model, and task.
OpenAI assures users will be able to see their shared balance and reset times in ChatGPT settings but has not detailed the level of granularity displayed. The company also has not answered questions about the exact size of this shared pool. This lack of figures fuels concerns, especially since OpenAI already reversed a capacity reduction deemed too aggressive for its $200 Pro offering at the end of October.
Un mouvement qui dépasse OpenAI
The convergence between chatbot and work agent is not isolated. In September, Anthropic merged Claude Chat and Cowork into a single interface, with the same idea: moving from simple questions to heavier tasks without switching tools. The trend is clear among major AI labs, which seek to make their products more continuous in use. One sensitive point remains: when functions converge, usages also mix, and the attention cost always ends up rising somewhere.
For OpenAI, the bet is twofold. On one hand, simplify the experience for subscribers and better integrate its products. On the other, avoid users seeing this unification as a disguised capacity reduction. At this stage, the balance will mainly depend on a very unglamorous but decisive detail: how many tokens the new quota will actually allow to consume, and how quickly it will be depleted.
Key points
- December: start of OpenAI quota unification.
- Plus, Pro, and Business are concerned.
- ChatGPT and Codex will share the same pool.
- Codex remains distinct on the desktop app, according to OpenAI.
- Anthropic has already merged Claude Chat and Cowork.
By the numbers
- 3 plans — Plus, Pro, and Business, affected starting December.
- $200 — price of the Pro subscription mentioned by sources.
- 5 hours — limit window still cited for some uses.
- July 2026 — launch of ChatGPT Work, according to sources.