THE ESSENTIALS IN 30 SECONDS

  • From December, Plus, Pro and Business subscribers will have one usage budget, measured in tokens, covering both ChatGPT and Codex, according to OpenAI's updated help page as reported by The Decoder.
  • Chats with Instant models will not count against the budget; heavier models and longer tasks will use it up faster, and users who hit the cap can wait, upgrade or buy extra credits.
  • OpenAI has not published an exact start date or the size of each plan's allowance.

OpenAI is changing how it rations usage on its paid ChatGPT plans. Starting in December, subscribers to Plus, Pro and Business will no longer have separate limits for ChatGPT and for Codex, the company's coding agent. Both will draw on a single allowance counted in tokens, according to an updated OpenAI support page described by The Decoder on November 1.

What is changing

Tokens are the units of text a model reads and writes, so the new budget depends on what people actually run. OpenAI's page says more capable models and more complex tasks consume the allowance faster. Conversations with the lighter Instant models are excluded, which means ordinary chatting keeps a flat-rate feel. Users will be able to see their remaining balance and reset times in Settings. Once the limit is reached, the options are to wait for the reset, move to a higher plan or buy additional credits.

The first public sign of the change came on October 29, when a notice in ChatGPT's checkout page, spotted by Tibor Blaho and reported by All-AI.de, said plan usage would have a shared limit across ChatGPT and Codex from December. Around the same time, OpenAI had doubled Work and Codex limits on several higher-priced tiers.

Separately, OpenAI is merging ChatGPT's Chat and Work modes, folding the agentic Work features into the main interface. The Decoder notes that users currently pick between two modes and eleven model settings, a complexity that has drawn criticism.

What is still unknown

OpenAI has not given an exact December date, the number of tokens each plan includes or the price of extra credits. Those details will decide how the change lands. The Decoder argues that it ends "all-you-can-prompt" pricing for heavy users of advanced models and agents, since a long Codex session and a demanding ChatGPT task will now compete for the same pool. Light users who mostly rely on Instant models may notice little difference.

The shift mirrors a wider move in the industry towards metered allowances for agentic work, which costs far more to serve than short chat replies.