Home / ChatGPT context window: size, limits and what to do when it fills up

ChatGPT context window: size, limits and what to do when it fills up

ChatGPT can only keep so much of a conversation in view at once. That limit is the context window, and ChatGPT never shows you how close you are. Here is what the numbers look like and how to stay ahead of them.

How big is it?

There is no single answer. The window depends on your plan and on which model answers, and OpenAI adjusts both. Third-party guides reviewed in 2026-10-08 report these figures for the ChatGPT app:

These are the app limits, not the API, which offers much larger windows. The full comparison is on context limits by plan.

What happens when a chat gets long

Each time you send a message, the whole conversation is sent along with it. As the chat grows, the model has more to weigh, and once it passes the window the oldest parts no longer fit. You may notice it:

• forgetting details or rules you gave at the start;
• contradicting something it said earlier;
• bringing back suggestions you already rejected;
• slower or less consistent replies.

These symptoms show up before a hard cutoff, which is why a meter helps: by the time you see the symptoms you have already lost time.

How to see how full your chat is

ChatGPT has no meter. You can paste the conversation into a token counter, or use ContextMeter, a free browser extension that shows the percent of the window used while you chat. It detects your plan from the sidebar, counts the whole chat, and turns amber and red as you approach the limit.

What to do when it is nearly full

1. Finish the current topic. Do not start something big.
2. Ask ChatGPT for a structured summary using the handoff prompt.
3. Open a new chat and paste the summary in.
4. Put anything you need in every chat into Custom Instructions or a Project, so it does not use up your window.

Frequently asked questions

How big is ChatGPT's context window?

It depends on your plan and the model. Third-party guides in 2026 report between 8K and 27K tokens for Free, 32K to 54K for Plus on standard models, and 128K for Pro, with larger windows for reasoning models. OpenAI changes these, so check its plan page.

Does ChatGPT tell me when the context window is full?

Not in the normal chat. There is no built-in meter, so you only notice through symptoms like forgotten details and repeated mistakes.

What happens when a ChatGPT chat gets too long?

The oldest parts of the conversation stop fitting. The model can ignore or lose early instructions and decisions, and answers can become less consistent and slower.

Does starting a new chat fix it?

Yes, a fresh chat starts with an empty window. You lose the earlier context, so carry over a summary of what matters.

How do I check how full my chat is?

ChatGPT does not show it. You can paste text into a token counter, or use an extension that counts the visible messages as you chat.

Last reviewed 2026-10-08. Numbers change often; see the sources linked on this page.