OpenAI to merge ChatGPT and Codex usage limits in December

OpenAI has announced a significant update to its subscription limits, combining the resource pools for ChatGPT web chat and Codex. The change is scheduled to take effect in December 2026 for Plus, Pro, and Business tier subscribers.

Previously, users could chat extensively in the browser and use Codex independently without one service depleting the limits of the other. Going forward, long conversations will consume the same pool as heavy coding tasks.

• ChatGPT and Codex limits will be merged into a single pool starting December 2026.
• Token consumption will replace simple message counting.
• The Instant mode is exempt and will not draw from the shared quota.
• Usage and reset times will be trackable in the account settings.

Transition to Token-Based Accounting Instead of counting individual messages, OpenAI will track the exact number of tokens processed and generated by the models. The resource consumption will depend directly on the chosen AI model and the complexity of the specific task.

This means a prolonged conversation using a resource-heavy model could potentially cost significantly more than dozens of simple queries. OpenAI has not yet clarified the exact accounting rules for advanced reasoning modes such as Medium, High, and Extra High.

Exceptions and Quota Management The Instant mode will remain an exception, allowing users to receive quick responses without consuming the shared resource pool. Account settings will feature a dedicated usage section to monitor remaining limits and refresh times.

Implications for Users and Developers This structural shift is driven by the gradual integration of ChatGPT Work agent capabilities into the main user interface. While OpenAI promises that most users will not experience more frequent limit exhaustion, developers using both chat and coding features will need to adapt to the new unified quota system.