OpenAI announced on its official doc page that it will make ChatGPT and Codex draw from 1 shared token allowance on Plus, Pro and Business plans.

The change starts in December, when OpenAI moves the agentic features of ChatGPT Work, its July agent built on Codex, into the main ChatGPT experience.

Work already shares Codex’s limits, while ordinary ChatGPT chats run on a separate meter, so heavy coding today leaves chat untouched.

After the switch, each request is priced by tokens. Currently ordinary ChatGPT chats your limit isn’t set by tokens. It’s set by message counts per model, such as a weekly cap on GPT-6 Pro messages