AI Token Counter

Paste a prompt to count its tokens, see how a model splits it up, and compare what it costs to send to any model in AI Benchmarks. It is free and runs in your browser.

Tokens, estimated
0
Characters
0
Words
0
Characters per token
0

Compare models for this text

0 input tokens and 500 output tokens, cheapest first. Choose a row to use that model in the estimate above, and add any model from AI Benchmarks below.

  1. Benchmarks
  2. Benchmarks
  3. Benchmarks
  4. Benchmarks
  5. Benchmarks
  6. Benchmarks
  7. Benchmarks
  8. Benchmarks
  9. Benchmarks
  10. Benchmarks
  11. Benchmarks
  12. Benchmarks
  13. Benchmarks
  14. Benchmarks
  15. Benchmarks
  16. Benchmarks
  17. Benchmarks
  18. Benchmarks
  19. Benchmarks
  20. Benchmarks

Type a model or maker, or use the arrow keys to browse.

Prices per million tokens from AI Benchmarks, updated November 4, 2026. Standard rates only: cached input, batch discounts and long-context surcharges are not included.

How your text splits into tokens

Each shaded piece is one estimated token. A leading space belongs to the word after it, just as it does in GPT-style tokenizers.

Your tokens appear here as you type. Try the sample text to see words, numbers, emoji and other languages split differently.

By type of content

  • English words0
  • Other languages0
  • Numbers0
  • Punctuation and symbols0
  • Spaces and line breaks0
  • Emoji0

What a token is

A token is the unit a language model reads and writes: a short word, part of a longer word, a group of digits, a punctuation mark or a piece of an emoji. Providers bill your prompt and the model's reply separately, and replies cost several times more per token, so a long answer can cost more than a long prompt.

  • Common English words are usually one token each, and longer or rarer words split into two or three.
  • Numbers are read in groups of up to three digits, so long IDs, prices and tables add up quickly.
  • Accented letters, non-Latin scripts and emoji take more tokens per character than plain English.
  • Asking for a shorter answer, or reusing a cached system prompt, is usually the quickest saving.
Hello world
Hello world2 tokens
tokenization
tokenization2 tokens
1234567890
12345678904 tokens
你好，世界
你好，世界About 5 tokens

Token counts on this page are estimates based on how GPT-style tokenizers split text. Claude, Gemini, Llama and other model families use their own tokenizers, so the same text can count a little higher or lower; use the provider's own token counting tool when you need an exact figure. Reasoning models also bill their thinking as output tokens, so allow extra output for them.

Good to know

How accurate is the token count?

It is an estimate. Each shaded piece in the token view is one estimated token, split the way GPT-style tokenizers split text.

Where do the model prices come from?

From the AI Benchmarks pricing data, updated November 4, 2026. The estimate uses standard rates per million input and output tokens.

Does the cost include cached input or batch discounts?

No. Only standard rates are used: cached input, batch discounts and long-context surcharges are not included.