← sources

ChatGPT gets metered in tokens in December. The habit I'm dropping first is keeping one giant chat for each project

source: reddit (reddit:OpenAI)published 2026-11-02 14:23ingested 2026-11-03 02:39

retained text

OpenAI's help page says that starting in December, Plus, Pro and Business usage will be shared across ChatGPT and Codex and measured in tokens, "the units of information a model processes and generates." Instant won't draw from your allowance. OpenAI also says it expects most users won't have to wait for resets more often than they do now, so this may not change much for you.

The page doesn't say how a long chat gets counted, and that's what I want to know. The model processes your earlier messages to answer the new one, at least the part that still fits in its context window. If those tokens count, message 300 in a giant chat costs far more than message 3 in a new one, even when the question is the same. OpenAI could discount that history. It hasn't said either way, so this part is my inference.

It wouldn't be unusual, though. Google's help page says outright that the same question uses more of your Gemini limit in a long chat, and Anthropic tells Claude users to start a fresh chat for a new topic.

So here's what I'm changing before December:

1. New chat when the task changes. Before I leave the old one, I paste this:

"I'm moving this work to a new chat. Write a handoff I can paste there: the goal, every decision we made and why, options we rejected so they don't get suggested again, any facts or numbers that matter, and the next step. Under 300 words."

1. Instant for anything that doesn't need thinking, like quick lookups and small rewrites. It's weaker, but it's the one option that doesn't touch the allowance.
2. No more asking the same thing twice. Finding an old answer costs no tokens. Asking again on a thinking model might cost plenty.
3. Check Settings > Usage once it's there. It's supposed to show your combined remaining usage and reset times. If you use Codex, your coding will draw from the same allowance as your chats.

I build AI Toolbox, a browser extension for ChatGPT, so I'm biased on the tools for this. It shows a free gauge on the page with an estimate of how full the current chat's context is, which helps you see when a chat has gotten big. That's not your allowance, which is what Settings > Usage is for. Its search finds words inside old chats (free shows the first 25 results, Premium shows all of them), and carrying a chat into a new one with a summary, like the prompt above, is Premium. The prompt works fine without it.

Has anyone seen OpenAI say how history in a long chat will count? And if you keep giant chats, are you planning to split them up or wait and see what December does?

document fetched 2026-11-03 00:28 · canonical https://reddit.com/r/OpenAI/comments/1xlyiu9/chatgpt_gets_metered_in_tokens_in_december_the