ChatGPT "Context window full": what to do

What "Context window full" means in ChatGPT

"Context window full" means the current ChatGPT conversation holds as much text as the model can use at once. It is a limit on this one chat. It is not a limit on your account. You can start a new chat right away, and the notice gives no reset time for the original conversation.

The context window is the amount of text a model can read at one time. In a chat, that is the conversation so far. OpenAI's Help Center article on this notice says it plainly: "Your messages and ChatGPT's replies both take up space, so a few long exchanges can fill a chat." When the conversation reaches the limit, ChatGPT shows the notice.

The exact wording, and the second notice

OpenAI's Help Center has an article for this notice, titled "Why does ChatGPT say "Context window full" or "Chat memory full"?". The wording is "Context window full", with a capital C and the other words in lowercase. "Chat memory full" is the companion notice. The Free Tier FAQ says both notices "refer to how much information ChatGPT can use in the current conversation." So the word "memory" in the second notice is about this chat. It is not about the saved Memory feature in your settings.

The help pages quote only these two strings. They do not quote the buttons under the notice, so the button labels in your app may read a little differently from the words used in this guide.

Who sees it

The first paragraph of OpenAI's article says: "On the Free plan, ChatGPT can show either notice when a conversation fills up." That sentence confirms that Free users can see it. It does not say that only Free users see it, and the article names no other plan. If you pay for ChatGPT and see the notice, the steps below still apply, because the pricing page lists a context window for every plan.

Context window full vs maximum length vs a usage limit

ChatGPT has an older message that reads "You've reached the maximum length for this conversation". The two strings are different. If that older one is on your screen, read what to do when ChatGPT says you've reached the maximum length for this conversation instead.

A usage limit is a different thing again. It caps how much you can send over a period of time, not how long one chat can grow. If your message is about how many messages you have sent, or about a limit that resets later, you hit a usage cap, not a context cap. Read what to do when ChatGPT says you have hit your limit for that case, and how ChatGPT usage limits work and when they reset for the timing.

The quick test is a new chat. OpenAI's FAQ says "You can start a new chat right away on the Free tier." A context limit belongs to one conversation, so a fresh conversation starts empty.

Memory settings do not fix it

The FAQ says you do not need to turn off Memory or delete saved memories, and you do not need to clear other chats. The limit is on the current conversation. Deleting saved memories or old chats does not free space inside it, so leave your settings alone.

What the automatic summary carries into a new chat

The notice may offer to continue in a new chat with a summary. OpenAI writes this with an "if": the option appears only when the notice offers it. Not every reader will see it.

The FAQ says the summary "can bring over some context". It is also clear about the cost: "It may leave out an earlier instruction or the exact wording of something you wrote." OpenAI's article on the notice says the same thing in fewer words: "A summary may leave out details." OpenAI does not say anything more about what the summary keeps.

Two kinds of work lose the most. The first is a set of instructions you gave at the start, such as a fixed format or a rule like "always answer in Spanish". The summary may keep the topic and drop the rule. The second is a draft. A summary may keep what your text is about and lose the exact sentences, which is a problem when the sentences are the work.

So continue with care:

• Reopen the original chat from your history. OpenAI's article says you can reopen it and copy over anything you still need, because the old chat stays there.
• Paste the passage you are working on. OpenAI's own advice is to "paste the passage you want to revise and tell ChatGPT where you'd like to pick up."
• Repeat the instructions that matter in your first message in the new chat. Keep them short.
• Read the first reply closely. If it breaks one of your rules, add that rule again before you go on.

What "upgrade" changes on each plan

Some notices also offer an upgrade. The FAQ says: "If the notice offers an upgrade, you can choose it to continue in the same chat." That is all OpenAI documents. Its help pages do not say which plan the upgrade moves you to, or how much room it adds.

What OpenAI does publish is the window size for each plan, on the ChatGPT pricing page. The page has one row for GPT Instant models and one for GPT Reasoning models. The figures below were read from that page on 7 November 2026.

[
  {
    "plan": "Free",
    "total_window_k": 27,
    "approx_input_pages": 12
  },
  {
    "plan": "Go",
    "total_window_k": 54,
    "approx_input_pages": 40
  },
  {
    "plan": "Plus",
    "total_window_k": 54,
    "approx_input_pages": 40
  },
  {
    "plan": "Pro",
    "total_window_k": 128,
    "approx_input_pages": 250
  }
]

The page writes the sizes as "27K" or "54K" and prints no unit. OpenAI's release notes measure the same kind of window in tokens. A token is a small piece of text, often part of a word. A footnote on the pricing page adds: "The portion available for user input is smaller than the total window." The reply needs room too. When OpenAI expanded the Thinking window on 20 February 2026, the release note gave the total as 256k tokens, split into 128k input and 128k max output.

The page also gives an input maximum in pages of text, shown in the approx_input_pages column. OpenAI writes these with a "~", so they are rough. Free gets about 12 pages in Instant and Pro about 250. OpenAI does not say how long a page is.

[
  {
    "plan": "Go",
    "total_window_k": 256,
    "approx_input_pages": 320
  },
  {
    "plan": "Plus",
    "total_window_k": 256,
    "approx_input_pages": 320
  },
  {
    "plan": "Pro",
    "total_window_k": 400,
    "approx_input_pages": 680
  }
]

Free is not in the second chart because its Reasoning rows read "Varies", with no number.

Here is what the two charts say about an upgrade:

• Moving from Free to Go doubles the Instant window, from 27K to 54K.
• Go and Plus show the same figures in both rows. On this page, moving from Go to Plus does not give you a larger window. Plus has other reasons to exist, which the ChatGPT Plus price and limits guide covers.
• Pro is the largest, at 128K for Instant and 400K for Reasoning. Pro is one column on the page, offered as "Your choice of 3 usage tiers", and the page does not say whether the window differs between those tiers.
• OpenAI's pricing page listed these prices on 7 November 2026: Go at $8 a month and Plus at $20 a month. Pro starts at $100, with "Plans at $100, $200, or $500 USD per month."

One more number causes confusion. OpenAI's models page says "GPT-6.1 Sol uses a default context window of 500,000 tokens in Codex and ChatGPT Work." That sentence is about Codex and ChatGPT Work. It is not a figure for ordinary chats on any plan, so do not expect a 500,000-token chat after an upgrade. For the models themselves, see how GPT-6 Astra, Sol and Luna differ in ChatGPT. If you are choosing a plan for reasons beyond the window, start with which ChatGPT plan you actually need.

What fills the ChatGPT context window fastest

Long pastes. Pasting a whole report, a long log or a full chapter is the quickest way to fill a chat. OpenAI changed how this works during 2026. In the 4 August 2026 release note, a paste of more than 10k characters became an attachment for Enterprise and Education, and the note says this was already available on all other plans. OpenAI says the change "helps prevent large pastes from consuming your entire context window". You can turn the attachment back into text with "Show in text field".

Read that sentence carefully. It says the change helps prevent a paste from taking the whole window. It does not say an attachment uses no space, and OpenAI does not say how much of an attachment counts. Treat a long paste as something that can still fill the chat.

Long replies. Your messages and the replies both take space. If you ask for a full rewrite of a long document on every turn, each turn adds a long reply to the chat. Ask for the changed section only, and the chat grows much more slowly.

Files and images. OpenAI's File uploads FAQ says "Text and document files are capped at 2 million tokens per file." That is an upload limit. It is not the size of your chat. It is far larger than any chat window on the pricing page, so an accepted upload tells you nothing about how much room is left. The same FAQ says "Content from a file that was brought into a conversation can remain in that conversation." OpenAI does not document how files or images count toward the context window.

How to fix "Context window full", step by step

Work through these in order. The first two cost nothing and work on every plan.

1. Take the summary if the notice offers it. Then repeat your key instructions in the first message of the new chat, because the summary may have dropped them.
2. Or start a clean chat from the sidebar and send only the question or passage you need. OpenAI suggests this when a summary chat fills up again.
3. Trim what you paste. Send the one section you want to change, not the whole document, and ask for replies that cover that section only.
4. Move reference material into a Project. OpenAI's Projects article says: "Add reference material to your project by uploading PDFs, spreadsheets, docs, images, or by pasting text." Project instructions apply inside that project, and "Chats can reference other conversations within the same project." A project holds up to 5 files on Free, 25 on Go and Plus, and 40 on Pro, Business, Enterprise and Edu, and you can upload 10 at once. OpenAI does not present Projects as a fix for this notice, and the Projects article does not say a project has a larger window. The practical gain is that your standing instructions and files live in the project, so you stop pasting them into every chat.
5. Try a Reasoning model in the new chat. On Go the pricing page lists 54K for Instant and 256K for Reasoning, and Pro shows the same pattern. OpenAI's articles on this notice do not mention switching model, so treat it as a comparison of the pricing rows, not as an official fix. On Free the Reasoning window reads "Varies", so a gain is not certain there.
6. Upgrade, if the notice offers it and the larger window is worth the price to you. Check the charts above first. Go to Plus shows no change in window size.

Why does a new chat fill up after only a few messages?

This one is different, because a short chat should not be anywhere near the limit. OpenAI's article gives these steps:

1. If you continued with a summary, choose New chat in the sidebar. Do not continue from the summary again.
2. Send a short test message in that separate chat. OpenAI's example is "What's the capital of France?".
3. If even the test shows the notice, check OpenAI's status page for an incident.
4. If there is no incident, open a private browser window, sign in to the same account, and try again.
5. If the notice still appears, contact OpenAI Support.

A one-line question in an empty chat is far below any window on the pricing page. That is why OpenAI's next steps after the test are the status page and Support, not trimming your text.

FAQ

Is "Context window full" only shown on the ChatGPT Free plan?

OpenAI does not say so. Its Help Center article says "On the Free plan, ChatGPT can show either notice when a conversation fills up," and names no other plan. That confirms Free users can see it, but it does not rule out paid plans. Every plan on the ChatGPT pricing page has a context window, so the same fixes apply wherever you see the notice.

Will deleting saved memories fix "Context window full"?

No. OpenAI's Free Tier FAQ says you do not need to turn off Memory or delete saved memories, and you do not need to clear other chats. "Context window full" and "Chat memory full" both refer to how much ChatGPT can use in the current conversation, so the fix is a new chat or a shorter one, not a change to your Memory settings.

What does the automatic summary leave out when I continue in a new chat?

OpenAI says the summary can bring over some context but "may leave out an earlier instruction or the exact wording of something you wrote." Repeat your important instructions in the first message of the new chat. If you are editing a draft, paste the passage you want to revise and say where to pick up. The original chat stays in your history, so you can copy text from it.

Does upgrading let me keep going in the same chat?

If the notice offers an upgrade, OpenAI says choosing it lets you continue in the same chat. OpenAI does not document how much room the upgrade adds. The pricing page on 7 November 2026 listed GPT Instant windows of 27K on Free, 54K on Go and Plus, and 128K on Pro, so going from Go to Plus shows no larger window.