Token counter and will it fit?
Paste text or open a file to see how many tokens it is, whether it fits each AI model in one go, and what sending it once costs. If it is too long, split it into parts you can paste one at a time. Nothing you paste leaves this browser.
Price a whole AI feature, with calls per month and the replies, in the AI Cost Calculator.
New to this? What it does, in plain words
AI models read text in tokens, pieces of words about four letters long. Every model has a limit on how many tokens it can read at once, and API prices are per token.
Paste a document, a transcript or a long prompt to see its size in tokens, words and pages, and whether each model can take it in one go. When something is too long for a chat app, split it into numbered parts that tell the AI to wait for all of them.
Exact or estimate
The count uses OpenAI’s own tokenizer, the one GPT-4o to GPT-5 use. Claude and Gemini use their own, so their numbers are marked ≈ until you count them exactly with a free key.
Where it goes
Nowhere. The counting happens in your browser, and files are read on your device.
Your text
The count
0tokens
- Claude
- ≈0
- Gemini
- ≈0
- Words
- 0
- Characters
- 0
- Pages
- 0
Counted with OpenAI’s o200k tokenizer, exact for GPT-4o to GPT-5. Numbers marked ≈ are estimates for other models.
| Model | Tokens | Share of what it reads at once | To send once |
|---|
The price is for sending the text as input once, at each model’s standard rate. Some models charge more above a size; the AI price index lists those rules.
Split it into parts
How tokens add up
A token is a piece of a word. Common short words are one token each; long, rare or foreign words break into several, and numbers, code and links use more tokens than ordinary prose.
- 1Count itPaste the text or open the file.
- 2Check the fitSee which models take it in one go, and what it costs.
- 3Split if neededNumbered parts that tell the AI to wait.
- 4Ask your questionOnce every part is in, ask about the whole.
Questions
How many tokens is 1,000 words?
About 1,300 to 1,400 tokens of ordinary English with OpenAI’s tokenizer, and somewhat more on Claude. Technical writing, code and other languages run higher. Paste your own text above for its real count.
Is this count right for Claude and Gemini?
Not exactly. Anthropic and Google don’t publish tokenizers that run in a browser, so their numbers here are estimates, marked ≈. Add a free Anthropic or Google AI Studio key in Settings and the page asks their own counting endpoint, which costs nothing and runs no model.
Why does ChatGPT say my message is too long when the model can read more?
Chat apps set their own limits on a single message, below what the model can read through the API, and they don’t always publish them. If a message is refused, split it into parts here and paste them one at a time; each part tells the assistant to wait until it has them all.
Does my text go anywhere?
No. The tokenizer runs in your browser and files are read on your device. The only exception is the optional exact Claude or Gemini count, which sends the text to that company with your own key when you press the button.