ChatGPT and Codex will share one usage allowance starting in December

What OpenAI announced

On October 31, OpenAI updated its Help Center article Understanding ChatGPT and Codex usage limits with a notice that starting in December, Plus, Pro, and Business plans' usage allowance 'will be shared across ChatGPT and Codex.' The company frames the move as preparation 'as we bring the agentic capabilities of Work into the main ChatGPT experience.'

Under the new system, usage for both ChatGPT and Codex 'will be measured in tokens, the units of information a model processes and generates.' The article says that 'more advanced models and complex tasks generally use more of your allowance,' and that 'Instant chat will not draw from your usage.'

Subscribers will see one combined remaining balance and its reset times in Settings, under Usage. At the limit, they can 'wait for it to reset, upgrade your plan, or buy credits to continue using eligible features.'

What already shares a pool

The shared allowance is not entirely new. OpenAI's existing guide to GPT-6 Astra in Work and Codex already says that 'Work and Codex share the usage allowance included with your plan.' Its Codex usage article adds ChatGPT for Excel and Workspace Agents to the same pool. What December adds is regular ChatGPT conversation — except chats using Instant models — to that same pool on Plus, Pro, and Business.

The October 31 article gives no start day beyond the month and no allowance sizes for any plan. OpenAI has released only one published yardstick: an estimated-messages-per-five-hour-window table for Plus and Standard Business, which SaasBuzz used to calculate that the same allowance covers roughly 25 messages with GPT-6 Astra and about 1,675 with GPT-6 Luna — a 67-fold gap set by model choice alone.

What this means for subscribers

For developers and teams who run Codex on Plus, Pro, or Business, the consequence is direct. Coding tokens and chat tokens come out of the same allowance. A heavy afternoon of agent work competes with every ordinary question asked in the chat window. Users who rely heavily on both chat and Codex may need to track their combined burn to avoid hitting limits mid-task.

The change arrives two days after OpenAI dropped a planned cut to ChatGPT Pro 200 and doubled the Work and Codex allowance on Pro 100 and Pro 500. That increase applied only to the agent and coding pool. From December, the same pool also pays for chat with every model except Instant.

Instant chat — OpenAI's fastest, cheapest tier — remains exempt. OpenAI did not specify which models count as Instant beyond the name, leaving ambiguity about how the boundary works in practice.

What OpenAI has not disclosed

OpenAI's announcement leaves several questions open: the day in December the change takes effect, how many tokens any plan includes, whether allowances grow to absorb the new chat burden, and whether the five-hour and weekly reset windows stay unchanged. The Astra guide says Pro plans currently have no five-hour limit in Work and Codex; it is unclear whether that carries forward.

The only reliable capacity-planning input is your own measured burn. OpenAI directs subscribers to Settings under Usage, where the combined balance and reset times will appear once the change is live.

Buying credits as an option

OpenAI's article lists buying credits as one of three options at the limit. The Help Center's credits guide says eligible Plus and Pro users can buy credits without changing plans, and that a negative balance can affect a later purchase. Credits are documented to cover usage on 'eligible features,' though OpenAI has not specified which features are ineligible.

For token-based Enterprise agreements, Codex usage is billed in US dollars instead of deducted from credits, with separate rates on a published card.

The broader product direction

The shared allowance is a product move as much as a billing one. OpenAI is knitting together ChatGPT and Codex under one interface and one meter, ahead of merging chat and Work into a single experience. The company has been testing this direction since it launched Work, and the October 31 article formalizes the meter change before the interface change.

For competing AI coding tools such as Anthropic's Claude Code and DeepSeek's Super-agent, the practical constraint of a shared chat-and-code allowance could push power users to consider alternatives that do not tax a single budget.