Pasting long text into AI chats

When a document won't fit in the message box, split it into numbered parts and send them in order — here's the reliable way to do it.

The problem

ChatGPT, Claude, Gemini and other AI assistants all cap how much you can put in one message. The exact limit varies by service, plan and model — and changes without notice — but the symptom is always the same: your paste gets cut off, rejected or silently truncated, and the AI answers based on half the text.

The fix: numbered parts with a "wait" instruction

Split the document into parts and send them one after another. Two details make this work well:

Number the parts so nothing is lost or reordered.
Tell the AI to wait until all parts have arrived before answering.

SplitMyText does both automatically. With Number Each Part switched on, every part starts with a header like [Part 2/5 — wait for all parts before replying], so the model holds its answer until you've pasted the last one.

How to do it

Open the splitter and paste your full text.
Set a part size comfortably under the chat's limit — if you don't know it, 8,000 characters is a safe, efficient default for the major assistants.
Turn on Number Each Part and click Split Text.
Use Copy Next to work through the parts in order — it tracks which one you're on.

A message to send first

The part headers already ask the AI to wait, but it helps to say what's coming before the first part. Send something like this, then start pasting:

I'm going to paste a long text in several numbered parts because it is too long for one message. Don't summarise or analyse anything yet. After each part, just reply "Got part N". I'll tell you when I've sent the last one.

After the final part, say so and ask your question: "That was the last part. Now summarise the whole text." Asking only once everything has arrived is what stops the AI answering from the first part alone.

How big should each part be?

There's no single right number, because every assistant counts differently and the limits move. A practical approach:

Start at 8,000 characters. If the chat rejects or truncates a part, halve it and split again — 4,000 or 3,000 fits almost anywhere.
Fewer, larger parts are easier for you and for the AI. Only go smaller when you have to.
AI limits are really counted in tokens, not characters. The token counter gives you an estimate for your text.

Long translations and transcripts

For translation work it matters where the cuts fall: a sentence split across two parts tends to be translated badly on both sides. The website cuts at word boundaries, so no word is ever broken. If you need each part to end on a complete sentence, or want the end of one part repeated at the start of the next so the AI keeps the context, those modes are in the SplitMyText app for iPhone, iPad and Mac and Android.

When the AI's reply is the part that's too long

The same trick works in reverse. If an assistant has written something too long to paste where you need it — a messaging app, a form, a forum post — you don't have to ask it to "send it again in two halves". Copy the whole reply, paste it into the splitter, set the limit of the place you're pasting into and copy the parts across one by one. The character limits table has the numbers for the common platforms.

Why use this tool for it

The text you're pasting into an AI chat is often exactly the text you don't want passing through yet another server. SplitMyText never uploads or stores anything — the splitting happens in your browser, and cuts land on word boundaries so no sentence is chopped mid-word. On a phone, the app does the same from the share sheet.