This free Claude token counter estimates how many tokens your text uses on today’s Claude models: Claude Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 5.5. Paste a prompt or a whole document to see the tokens, what it costs on each model and how much of Claude’s 1M-token context window it fills.
Claude counts text differently from ChatGPT and from older Claude models, so a generic token counter can be off by a third or more. This one uses the ratios Anthropic publishes for the current Claude tokenizer.
Free Claude token counter
How many Claude tokens is your text?
Paste anything to estimate its tokens on today's Claude models, using Anthropic's own published ratios. You also see what it costs on each model and how much of Claude's context window it fills.
Cost on each current Claude model
| Model | Input price | Output price | This text |
|---|
Compare against older Claude models too
Unlock the side-by-side count for Claude models before Opus 4.7, so you can see how much more your prompts cost after an upgrade.
Unlocked: the older-model count is in the panel above. You are also on the weekly HOW TO // AI newsletter.
Free. You also get the HOW TO // AI newsletter once a week. Unsubscribe any time.
How we estimate: Anthropic says 1M tokens is roughly 555,000 words or 2.5 million characters on the current tokenizer (introduced with Claude Opus 4.7), and that older models fit about 750,000 words in 1M tokens. We average the word-based and character-based estimates. Prices are base API prices per million tokens; Batch is 50% off. For an exact count, use Anthropic's free token counting endpoint. Sources: Models overview, Token counting, Pricing, checked October 18, 2026.
How to use it
- Paste your text. The token, word and character counts update as you type, and nothing leaves your browser.
- Enter how long you expect Claude’s reply to be, so the cost includes output tokens.
- Check the table for the cost on each Claude model, and the meter for how much of the context window you’re using.
How Claude tokens compare to words
Anthropic says 1 million tokens is roughly 555,000 words, or about 2.5 million characters, on the tokenizer used since Claude Opus 4.7. Models before that fit about 750,000 words in the same million tokens, so the same text now counts about 30% more tokens.
| Claude tokens | About this many English words (current models) | On models before Opus 4.7 |
|---|---|---|
| 1,000 | 555 | 750 |
| 10,000 | 5,550 | 7,500 |
| 200,000 | 111,000 | 150,000 |
| 1,000,000 | 555,000 | 750,000 |
Claude context window and output limits
Every current Claude model has a 1M-token context window and can write up to 128K tokens in one reply on the standard Messages API. Your prompt, any documents and Claude’s reply all share the context window, so leave room for the answer.
How to get an exact Claude token count
This page estimates. For a closer number, Anthropic’s token counting endpoint returns the count for any current model before you send a request. Anthropic notes it can differ slightly from the final count, and it’s free to use (with rate limits). Developers can call it with the same messages they plan to send.
Related on HOW TO // AI
- AI token calculator: estimates for ChatGPT and other models too.
- Claude pricing: every Claude plan for people and teams, with usage limits.
- Claude Pro vs Max: which paid plan fits how much you use Claude.
Checked against Anthropic’s models overview, token counting and pricing documentation on October 18, 2026.
Frequently asked questions
How many words is a Claude token?
On current Claude models, Anthropic says 1 million tokens is roughly 555,000 words, so 1,000 tokens is about 555 English words. Models before Claude Opus 4.7 fit about 750 words per 1,000 tokens.
Why do newer Claude models use more tokens?
Claude Opus 4.7 introduced a new tokenizer. Anthropic says the same text produces about 30 percent more tokens than on earlier models, so recount prompts on the model you plan to use.
What is Claude's context window?
All current Claude models (Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 5.5) have a 1M-token context window and up to 128K output tokens per reply on the standard Messages API.
How do I get an exact Claude token count?
Use Anthropic's token counting endpoint, which returns the input token count for a given model before you send the request. Anthropic calls it an estimate that can differ slightly from the final count. It is free to use, subject to rate limits.