AI text and context-size estimator
Measure pasted text or a local text file, then compare a rough token range with a budget you enter. Exact text measurements help you trim a source; the token range helps you plan, but it cannot verify a model limit.
Text stays in this browser. File limit 1 MB; text limit 200,000 characters.
Paste text, choose a UTF-8 text file, or try the example.
Exact text measurements
- Characters (Unicode code points): 0
- Words (non-whitespace groups): 0
- UTF-8 bytes: 0
- Paragraphs (blank-line groups): 0
Approximate token range
About 0–0 tokens, using 4 characters per token ±1. This is a rough text-only heuristic; model tokenizers, languages and request formatting can change the real count.
How the estimate works
Choose a characters-per-token assumption from 2 to 12. The displayed range divides Unicode character count by one character above and below that assumption, rounding each end up. The default assumption is 4, so the range uses 3 to 5 characters per token. It is a heuristic for text only, with no model-specific tokenizer or request overhead.
Enter your own positive whole-number token budget to see a percentage range. This tool does not name a model, estimate price, or guarantee a fit. Check the complete request with the relevant model tokenizer or API when an exact count matters.
Frequently asked questions
Is this an exact token counter?
No. The token range is an estimate from a visible characters-per-token assumption. An exact count depends on the chosen model tokenizer and the complete request.
How are the exact measurements counted?
Characters are Unicode code points, words are non-whitespace groups, bytes use UTF-8, and paragraphs are blocks separated by blank lines. These rules are shown beside the result.
Are my text and file sent to a server?
No. The analysis runs in your browser. Choose a UTF-8 text file up to 1 MB or paste up to 200,000 characters.