Text splitter

How it works

1. 1

   Paste your text

   A long article, transcript, chat export, contract or notes. Anything you can copy, up to about 3 million characters.

2. 2

   Pick a limit

   Choose tokens, words or characters, set the most each chunk may hold, and add overlap if you want each chunk to repeat the end of the one before.

3. 3

   Copy or download the parts

   Each chunk shows its exact size. Copy them one at a time, or download every part in a single .txt file. Nothing is uploaded or stored.

Why split text at all?

Every AI chat has a cap on how much you can paste in one message, and every model has a context window it can hold at once. Search and RAG systems work best on small, focused chunks rather than one huge block. Splitting by hand means counting, guessing and cutting sentences in half. This splitter measures each chunk exactly and ends it at a natural break, so each part reads cleanly on its own.

A fixed count, such as one chunk every 4,000 characters, is quick but slices through sentences and headings. Breaking at paragraphs first keeps related lines together, and chunks come out a little uneven in size, which is normal and better for meaning.

Choosing a chunk size

For a search or RAG index, 500 to 1,000 tokens per chunk with a little overlap is a common place to start: big enough to hold a full idea, small enough to match a specific question. For pasting into a chat, larger parts mean fewer messages; pick a size well inside what the chat accepts. As a rough guide, 1,000 tokens is about 750 English words.

Related free tools

This splitter works on pasted text. To start from a PDF, turn it into text with PDF to Markdown, or let the free knowledge base builder clean and chunk the file for you. The token counter shows a text's total size against each model's limits, and the PDF word counter sizes up PDFs before you convert them.

Frequently asked questions

How do I split text into chunks?

Paste the text into the box, choose whether to measure chunks in tokens, words or characters, and set the largest size a chunk may be. The text is split as you type, and each part gets its own Copy button. Download all as .txt saves every part in one file, with a divider between parts.

Where does the splitter cut the text?

It never cuts in the middle of a word unless a single word is longer than the limit, such as a very long link or text written without spaces. Within the back half of each chunk it looks for the best place to stop: a blank line between paragraphs first, then a line break, then the end of a sentence, and only then an ordinary space. Each chunk is an exact piece of what you pasted, so line breaks and formatting inside it are kept.

Should I split by tokens, words or characters?

Use tokens when the limit you are working around is an AI model's, since models count tokens rather than words. Token counts here are exact for current OpenAI models (the o200k_base encoding used by GPT-4o, GPT-4.1, the o-series and GPT-5). Claude and Gemini use their own tokenizers, so for them treat the count as a close guide and leave some headroom. Use words or characters when a form, a database field or a style guide sets the limit.

What is chunk overlap, and do I need it?

Overlap repeats the last part of each chunk at the start of the next, so a sentence or idea that sits on a boundary appears whole in at least one chunk. It is common when building a search or RAG index, where something like 10 to 15 percent overlap (50 to 150 tokens on 1,000-token chunks) is a typical starting point. If you are pasting parts into a chat one after another, leave overlap at 0 so nothing is repeated.

How do I paste a long text into ChatGPT or Claude in parts?

Choose the Labels + "wait for all parts" note option. Each part starts with [Part 1 of 5] and a short note asking the assistant to only acknowledge it until the last part arrives, so it doesn't start answering halfway through. Send the parts in order, then ask your question after the last one. Very long texts can still be more than a chat can hold at once, so check the total with the token counter first.

Is my text uploaded anywhere?

No. Splitting happens inside your browser tab. When you split by tokens, the page downloads the tokenizer once, but your text itself is never sent to our servers or anyone else's, so it is fine for private documents.

More free tools

All free tools →

Related guides