ChatGPT and Codex will share usage from December; Instant chat is excluded

OpenAI says ChatGPT and Codex will share a usage allowance from December for Plus, Pro and Business subscribers. Instant chat will be excluded. People who move between advanced conversations and coding will therefore have a combined balance to watch, according to an official help article checked on November 1.

The article says usage will be measured in tokens, with consumption varying by model and task complexity. It gives neither a precise December start date nor the size of the new allowances, leaving the effect on individual subscribers unresolved.

A publicly reproduced OpenAI subscriber email also describes a small-group test combining Chat and Work over the coming weeks. BleepingComputer published the copy on November 1. The test would let an idea discussed in conversation proceed into a task within the same experience.

The email gives examples involving files and plugins: building a website, creating a presentation or completing work on the web while the user reviews and guides it. Codex will remain a separate experience in the ChatGPT desktop app.

A broader pool than today’s agent allowance

Work and Codex already share an agent allowance under existing documentation. Eligible features such as Excel and Workspace Agents can use that pool too. Enterprise and Edu usage views exclude regular Chat. The announced transition extends shared accounting as agent capabilities move into the main conversation experience.

Tokens include material the model reads as well as what it writes: prompts, files, conversation history and tool results. Pricing documentation says context, reasoning, tools and caching affect consumption. At present, Fast mode spends included subscription allowance at 2.5 times the Standard rate for the same model. Those are current Work and Codex terms, rather than a published December capacity comparison.

Existing guidance separately describes ChatGPT file-upload, image and voice caps, each with its own resets and notices. Images generated in Codex count toward its general task allowance. The December announcement does not resolve every interaction between these feature-specific limits and the future shared pool.

Users have asked for clarification in Codex’s public issue tracker. An October 31 Plus subscriber request asks OpenAI to preserve separate chat and coding allowances. A different November 1 post describes using web conversations for scientific or engineering planning before implementing the result in Codex. These are requests and concerns, not measurements of the future policy’s effects.

Another Plus subscriber’s November 1 post asks for a before-and-after comparison of effective capacity. It requests clarity on advanced-model Study Mode and file analysis, and on how learning and coding would compete for allowance. OpenAI says it expects most users will not need to wait for resets more often than they do now.

Credits provide paid continuation for eligible features after included usage is exhausted. OpenAI’s personal-plan guide says the included allowance is consumed first, then available credits. Purchased credits are valid for 12 months; purchase options and availability vary by account, region and plan. These are separate from developer API credits.

Existing Codex guidance says an active turn can continue after the allowance is exhausted, subject to fair-use limits. After that turn, the account displays available options, such as credits, an eligible reset, an upgrade or waiting. The same guide says Support does not reset usage limits.