AI Tool Signals
๐Ÿงฎ Free tool

Token Counter

Paste any text to see its exact token count for GPT-4o, GPT-4 and GPT-3.5, plus words, characters and how much of each model's context window it uses. It runs in your browser, so nothing you paste is ever uploaded.

Model family

GPT-4o, GPT-4.1, GPT-5, o-series

0tokens
0words
0characters
0sentences

Context window used

GPT-4o / GPT-4.10.0%
Claude (Sonnet/Opus)0.0%
Gemini 1.5/2 Pro0.0%

Counted in your browser. Your text never leaves this page.

What a Token Actually Is

AI models do not read words, they read tokens. A token is a short sequence of characters, often a fragment of a word, that the model treats as one unit. Common words are a single token, while longer or rarer words split into several. As a working rule, one token is about four characters of English, and 100 tokens is roughly 75 words, but the only way to know for sure is to count the exact text, which is what the tool above does.

The count matters because tokens are the currency of every large language model. You are billed per token on the API, every model caps how many tokens it can read at once, and an over-long prompt is silently truncated or rejected. Counting first is how you avoid a surprise bill or a failed call.

Why Token Count Decides Your AI Bill

Every major AI API prices by the token, split into input and output. A prompt that looks short can be more tokens than you think once you include a system prompt, examples and context. Knowing the number before you send it is the difference between a predictable bill and a nasty one, especially at volume.

If you are weighing up which paid assistant to run day to day rather than calling the API, our ChatGPT review and our ChatGPT vs Claude comparison break down what each plan includes and where the real limits sit.

Which Tokenizer Matches Your Model

OpenAI has used two main tokenizers in recent years, and picking the right one changes the count. o200k_base is the newer, more efficient encoding behind GPT-4o, GPT-4.1, GPT-5 and the o-series. cl100k_baseis the older one still used by GPT-4, GPT-4 Turbo and GPT-3.5 Turbo. The toggle above switches between them, and for the same text the newer encoding usually returns fewer tokens.

Claude and Gemini use their own tokenizers that are not published as browser libraries, so this tool does not claim an exact count for them. The context-window bars give a close estimate using the OpenAI count, which is accurate enough to tell whether your text will fit.

How Many Tokens Can Each Model Take?

The context window is the total tokens a model can consider at once, your input plus its output. GPT-4o and GPT-4 Turbo handle about 128,000 tokens, Claude models around 200,000, and Gemini 1.5 and 2 Pro stretch to a million or more. A long document can fit comfortably in one, yet overflow another, which is why the tool shows the percentage used for each. For a wider view of the assistants themselves, see our best AI chatbots ranking.

Frequently Asked Questions

What is a token?

A token is the unit an AI model reads text in. It is usually a short chunk of a word rather than a whole word, so "tokenizer" might be two or three tokens. As a rough guide, one token is about four characters of English, and 100 tokens is roughly 75 words.

How many tokens is 1,000 words?

Around 1,300 to 1,500 tokens for typical English text, though it varies with punctuation, formatting and language. Code and non-English text usually use more tokens per word. Paste your exact text above for the real number rather than relying on the estimate.

Is this token counter accurate?

For OpenAI models, yes. It uses the same byte-pair encodings OpenAI uses: o200k_base for GPT-4o, GPT-4.1, GPT-5 and the o-series, and cl100k_base for GPT-4, GPT-4 Turbo and GPT-3.5. Claude and Gemini use their own tokenizers that are not published for the browser, so treat their context-window percentages as close estimates.

Does my text get uploaded anywhere?

No. The tokenizer runs entirely in your browser using JavaScript. Your text is never sent to our servers or anyone else's, which makes it safe to paste confidential prompts, drafts or client material.

Why do I need to count tokens?

Three reasons. API pricing is per token, so token count is your cost. Every model has a context window measured in tokens, so you need to know if your text fits. And prompts that run long get truncated or rejected, so counting first saves failed calls.

What is the difference between o200k and cl100k?

They are two different tokenizers. o200k_base is the newer one used by GPT-4o and everything after it, and it is more efficient, meaning fewer tokens for the same text. cl100k_base is the older encoding still used by GPT-4, GPT-4 Turbo and GPT-3.5 Turbo. Pick the one that matches the model you are using.

How many tokens can I send at once?

It depends on the model's context window. GPT-4o and GPT-4 Turbo handle about 128,000 tokens, Claude models handle around 200,000, and Gemini 1.5 and 2 Pro reach up to a million or more. The bars above show how much of each window your text uses.

More tools and reviews on AI Tool Signals. We test AI tools on real work and publish what they actually cost.

Independent reviews ยท affiliate disclosure
Token Counter: Free GPT & Claude Token Calculator