Byte Size Calculator

What the nine lines count

UTF-8 bytes is the length of a TextEncoder encoding of the box. ASCII stays one byte. é is two. A typical emoji such as 😀 is four UTF-8 bytes, four UTF-16 bytes, and one character. Characters walks the string by Unicode code point, so that emoji counts once even though it occupies a surrogate pair. ASCII characters counts those code points whose first code unit is below 128. The emoji's first code unit is a surrogate, which is above 127, so it lands on Non-ASCII chars and ASCII characters stays 0. A box of only café is Characters 4, ASCII characters 3, Non-ASCII chars 1.

UTF-16 bytes is the JavaScript string length multiplied by 2. That length is code units, not code points, so the emoji contributes 4 and é contributes 2. Kilobytes (UTF-8) divides the UTF-8 count by 1,024 and prints four decimal places. Megabytes (UTF-8) divides by 1,024 × 1,024 and also prints four places, which is why 5 bytes shows as 0.0000 MB. The grouping on the byte numbers uses en-US separators, so 4096 on the budget lines is written 4,096. This page does not offer a decimal SI kilobyte of 1,000.

The cookie line, the header line, and Copy output

Both budget lines use the UTF-8 count, not the UTF-16 count and not the character count. Cookie (UTF-8) compares that count with 4,096. The comparison is inclusive: 4,096 bytes of ASCII still says fits one cookie, and 4,097 says exceeds one cookie. Header (UTF-8) uses 8,192 the same way. A paste of 5,000 c characters exceeds the cookie line and still fits the header line. A paste of 9,000 exceeds both. The wording is fits one cookie, exceeds one cookie, fits one header field, or exceeds one header field, always after the phrase N B of 4,096 B or N B of 8,192 B.

The 4,096 figure is the size browsers are expected to support for one cookie's name, value, and attributes together. Paste the pair you will actually store, not a prose description of it. The 8,192 figure is one buffer in nginx's default large_client_header_buffers. Another server, or a different nginx setting, can cut off sooner or later. The page does not parse Set-Cookie attributes, and it does not add the overhead of a header name you left out of the box. Copy output copies the padded lines. There is no control that copies only the UTF-8 integer.

When the limit is bytes

Use this page when the limit is storage or a protocol budget: a JSON body, a cookie, a single header field, or a string a database will store as UTF-8. A character count answers a different question. The word counter reports characters as UTF-16 code units, including spaces, and it never prints a byte size. café is 4 characters there and 5 UTF-8 bytes here. An emoji that is one character on this page can still be two code units on the word counter. Reading time and sentence counts stay on that page.

A hex listing of the same bytes is the hex encoder, which writes each UTF-8 byte as a lowercase pair. This page will not turn café into c3 a9. A typed quantity such as 5 megabytes, converted between decimal and binary prefixes, is the unit converter. This page will not take the number 5 and emit kibibytes. It only measures the paste. A Flesch score and a top-words list are on text statistics, which also does not encode the draft.

Emoji, an empty box, and the exact 4,096 boundary

An empty box is a valid measurement. Every count is 0, kilobytes and megabytes print 0.0000, and both budgets say fits. Clearing the box returns that report. A single space is 1 UTF-8 byte, 2 UTF-16 bytes, 1 character, and 1 ASCII character. A newline is one byte in UTF-8 and one code unit in UTF-16, so it adds 1 and 2. The page does not strip a trailing newline you pasted from an editor. If the cookie line flipped to exceeds while the text still looks short, look for an emoji, a non-ASCII letter, or a newline you cannot see. é costs two UTF-8 bytes. 😀 costs four.

The character count and the UTF-16 byte count diverge exactly when a code point needs a surrogate pair. 😀 is Characters 1 and UTF-16 bytes 4. Two emoji joined by a zero-width joiner are more than one character if the spread produces more than one code point, and the UTF-8 line follows the encoder, not the grapheme you see. Combining marks add code points. The page does not normalize to NFC before counting, so é written as e plus a combining acute is more bytes than the single character é. Paste the bytes you will send. Copy output is the nine lines with their padding spaces, not a JSON object.

Rules and limitations

UTF-8 bytes: TextEncoder length. UTF-16 bytes: JavaScript string length × 2. Characters: code points. ASCII characters: code points whose first code unit is below 128. Kilobytes and megabytes: UTF-8 ÷ 1,024 and ÷ 1,048,576, four decimal places.

Cookie budget: 4,096 UTF-8 bytes, inclusive. Header budget: 8,192 UTF-8 bytes, inclusive. The lines say fits or exceeds. Copy output copies the nine padded lines. There is no download and no second encoding.

Common problems

If the character count and the UTF-8 count disagree, the extra bytes are non-ASCII. é is 2, and a typical emoji is 4. The word counter's character figure is code units, not this UTF-8 line.

If a cookie you believe is under 4 KB is marked exceeds, the box includes attributes or the name, or the value has characters outside ASCII. The header line is one 8,192-byte buffer, not every server's limit.

Frequently asked questions

About this tool

Maintained by Toolbench. For large or deeply nested files, split the input into smaller batches and verify representative results before using them in an import or application. Report reproducible problems through Contact.