ChatGPT and Codex will share one usage pool in December

OpenAI will put Plus, Pro and Business subscribers on one token-metered allowance shared by ChatGPT and Codex starting in December, the company said in a help center article from OpenAI on November 1. On a plan used for both, work on advanced ChatGPT models and Codex sessions will draw down the same balance, while Instant chat, which OpenAI calls “unlimited”, stays outside it.

The shared meter arrives with a product change. In an email to subscribers whose text was posted on Hacker News, OpenAI says that in the coming weeks ChatGPT will take on jobs like building sites and presentations or taking action on the web without a switch to the Work tab. ChatGPT “can decide when to use these capabilities” and breaks complex tasks into smaller steps, so a request typed into the main chat can become multi-step work charged to the allowance Codex also uses.

How much each request costs will depend on the model and the task, with more advanced models and complex tasks using more of the allowance. OpenAI says it expects most users won’t have to wait for a reset more often than they do now. Its own Codex estimates already vary widely: on Plus, it puts the allowance at 15 to 160 local GPT-6.1 Sol messages per five-hour window, or 350 to 3,000 on GPT-6 Luna, so the Sol range alone spans more than tenfold.

The combined remaining balance and reset times will appear in Settings → Usage. Once the allowance runs out, the choices are to wait for the reset, upgrade the plan or buy credits to keep using eligible features.