OpenAI will combine usage allowances across ChatGPT and Codex starting in December, affecting Plus, Pro, and Business subscribers, according to a help article the company published over the weekend. Under the new approach, conversations in ChatGPT and tasks in Codex will pull from a single shared pool of tokens, with more resource-intensive models and complicated tasks eating up a larger portion of the allowance. The company hasn't yet disclosed how large the combined allowance will be, though it says instant chat—ChatGPT's quicker-response mode for routine queries—will remain exempt from the shared limit.
The shift represents OpenAI's latest move to unify its expanding portfolio of AI products under one umbrella. In July, the company rolled out ChatGPT Work, an agentic tool powered by its Codex coding technology that lets users hand off complex assignments such as topic research, data analysis, and creation of documents, presentations, and spreadsheets. That launch also saw OpenAI merge ChatGPT and Codex into one desktop application with separate interfaces for conversations, agentic work, and coding. Right now, Codex, ChatGPT Work, ChatGPT for Excel, and Workspace Agents already draw from a shared agentic allowance, while standard ChatGPT conversations have their own separate usage caps. It's that division that will disappear in December, putting users in a situation where heavy Codex use cuts into their ChatGPT conversation allowance, and the other way around.
The decision to merge usage limits has drawn pushback from paying subscribers worried they'll end up with less access to the tools they're already covering. In a discussion thread launched on OpenAI's developer forum over the weekend, users urged the company to either scrap the consolidation or promise that subscribers won't wind up with reduced usable capacity under the new arrangement. "ChatGPT and Codex serve different purposes, and their usage limits should remain separate," one forum member wrote. Another user called for more granular usage reporting, including a breakdown of consumption by product, model, and task, adding, "Without that transparency, I will consider the change a downgrade to my paid plan and reconsider renewing." OpenAI confirmed that subscribers will be able to view their combined remaining allowance and reset times in ChatGPT's settings, but hasn't clarified what level of detail that will include.
The change arrives just days after OpenAI walked back a controversial plan to cut usage limits in half for its $200 Pro subscription. On October 29, the company cancelled the Pro 200 reduction, restoring higher Work and Codex allowances for subscribers already moved to the lower caps, while doubling allowances for Pro 100, Pro 500, and Business Premium tiers. But one forum participant questioned whether those increases would actually benefit subscribers once ChatGPT conversations start drawing from the same pool as Codex. "Doubling a resource allocation while simultaneously expanding the number of things consuming it doesn't inherently translate into more usable capacity," they wrote. OpenAI says it expects most users to hit usage resets no more often than they do now, but without publishing the new token allowances or spelling out how they stack up against existing limits, subscribers have little way of verifying whether that assurance will hold for their particular use patterns. The New Stack contacted OpenAI to ask when exactly the changes would roll out in December and whether the new combined allowance would match the existing separate allowances or leave some subscribers with less capacity, but the company declined to answer, instead pointing to its recent usage limit increases and existing documentation. The consolidation mirrors a broader industry pattern, with Anthropic merging Claude Chat and Cowork into a single interface in September, signaling that leading AI labs are pursuing a common vision of letting users switch seamlessly between asking questions and delegating more substantial tasks. Yet the execution of that vision—particularly around how usage is metered and communicated—may determine whether subscribers view the integration as a convenience or a constraint.

