ChatGPT Character Limit: What Actually Limits Your Text
ChatGPT doesn't have a single fixed character limit that applies everywhere. What limits you is measured in tokens, not characters, and the ceiling depends on the model you are using, your plan, and whether you are in the chat app or using the API. In practice you hit two separate limits: how much text you can send in one message, and how much the model can keep in mind across a whole conversation. Both have grown over time and will likely keep changing, so check the current documentation for exact figures.
This guide explains how those limits work, what happens when you run into them, and how to work with long documents without losing quality.
Characters vs tokens
Language models don't read text letter by letter. They split it into tokens, which are chunks that can be a whole short word, part of a longer word, a punctuation mark, or a space plus a word. As a rough rule of thumb often quoted for English, one token averages around four characters, or about three quarters of a word. That ratio shifts with the content:
- Plain English prose is fairly efficient.
- Code, unusual names, URLs and long numbers break into more tokens.
- Many non-English languages, especially those using non-Latin scripts, can use noticeably more tokens for the same meaning.
So two pieces of text with the same character count can use very different numbers of tokens. That is why character counts are only a loose guide. A word counter gives you a quick sense of size, but treat any token estimate based on it as approximate.
The three limits you can run into
1. Input per message
There is a practical cap on how much you can paste into a single message. If you exceed it, the interface may refuse the message or warn you that it is too long. The cap is generous for most everyday tasks but can be reached with long reports, transcripts or book chapters.
2. The context window
The context window is the total amount of text the model can consider at once: your current message, earlier messages in the conversation, any uploaded file content it is using, and its own replies. When a conversation grows beyond that window, the oldest material effectively drops out of view. You won't always get a warning. The symptom is the model "forgetting" instructions or details from early in the chat.
3. Output length
Responses also have a maximum length. For long requests, the model may stop mid-answer, or wrap things up more briefly than you wanted. Separately, models often choose to be concise even when they could write more, which feels like a limit even when it isn't one.
Signs you've hit a limit
- An error saying the message is too long.
- A response that cuts off mid-sentence.
- The model ignoring rules you set at the start of a long chat.
- Summaries of a long document that skip whole sections, especially the middle.
- Contradictions with things it said earlier.
How to work with long documents
Split the text into sections
Break the document at natural boundaries: chapters, headings, or topic changes. Process each section separately with the same instructions, then combine the results. Label each chunk clearly ("Part 3 of 7: Methods") so the model knows where it is.
Summarise in layers
For very long material, summarise each section first, then ask for a summary of the summaries. This keeps detail from earlier sections from getting crowded out. Inkwell's text summarizer works the same way on a single block of text and is useful for condensing sections before you combine them.
Restate key instructions
In long conversations, repeat the essentials every so often: "Remember, British spelling, no bullet points, keep each answer under 200 words." It costs a few tokens and saves a lot of drift.
Start a fresh chat when the topic shifts
A new conversation starts with an empty context. If you have spent forty messages on one draft and want to switch to something else, open a new chat and paste in only what matters.
Ask for long outputs in parts
Instead of "write a 5,000-word guide," ask for an outline first, then request each section on its own. You get more consistent length and you can correct direction early.
Use "continue" carefully
If a response cuts off, asking it to continue usually works, but check the seam. The model may repeat a sentence or change format slightly where it picks up.
Cutting text down to fit
Sometimes the simplest fix is to make your input shorter:
- Remove boilerplate such as repeated headers, footers, legal disclaimers and navigation text from copied web pages.
- Delete sections irrelevant to your question. If you only need feedback on the conclusion, don't paste the whole thesis.
- Convert tables to concise lists if formatting is bloated.
- Summarise background material in a few sentences rather than pasting it in full.
A text shortener can trim wordy passages while keeping the main points, which helps when you need to fit supporting material into a prompt.
Limits on other platforms
Character limits you meet around ChatGPT are often not ChatGPT's at all. Social platforms cap post length, ad platforms limit headline characters, and SMS has its own segment sizes. If you are asking the model to write for one of these, state the limit in your prompt ("under 280 characters including spaces") and check the result yourself. Models are not reliable at counting characters precisely, because they process tokens rather than individual letters. Expect to trim by hand or run the output through a counter.
The API is different
Developers using the API see token limits more directly. Each model has a documented context length and maximum output tokens, and requests are billed by tokens in and out. If you are building something on the API, read the current model documentation rather than relying on numbers quoted in blog posts, since these change as new models are released.
Practical prompt examples
I'm going to send you a long report in 5 parts. After each part, reply only with "Received part X." After part 5, I'll give you instructions.
Here is section 2 of 6. Summarise it in 150 words, keeping all figures exactly as written.
Rewrite this paragraph to be under 600 characters including spaces. Then tell me which ideas you removed.
The first prompt is useful for staging, but remember it doesn't expand the context window. If the total is too large, the earliest parts may still drop out of view.
FAQ
What is the ChatGPT character limit per message?
There is no single published character figure that applies to every model and plan, and the limits are measured in tokens. If a message is too long, the app will tell you. For very long text, split it into sections.
Why does ChatGPT forget what I said earlier?
Long conversations can exceed the model's context window, so the earliest messages fall out of view. Restating key instructions or starting a new chat with a short summary fixes most cases.
Can ChatGPT count characters accurately?
Not reliably. It processes tokens, not letters, so character counts in its answers are often a little off. Use a counter for anything with a strict limit.
How do I get ChatGPT to write longer responses?
Ask for the content in sections, give a target length for each, and build from an outline. One very long request is more likely to come back shortened or cut off.