Skip to content
The Internet Compass

AI Tools

AI Token Counter

Word count is a poor proxy for tokens — punctuation, whitespace and non-English text all tokenize differently. This runs a real BPE tokenizer client-side.

FreeNo sign-upRuns in your browser
Tokens
23
Characters
124
Words
18
Chars per token (avg)
5.39

How it works

  1. 1Paste your text
  2. 2It's tokenized using the same byte-pair-encoding scheme modern models use
  3. 3See exact token, character and word counts

Who this is for

  • Sizing a prompt against a model's context window
  • Estimating API cost before sending a large request
  • Understanding why a prompt costs more than its word count suggests

Frequently asked questions

Does this match every model's exact tokenizer?
It uses a widely-used BPE scheme (the o200k/cl100k family). Token counts are typically within a few percent across major model families, but not bit-for-bit identical for every provider.