Base64 Encoder / Decoder

Written byOleh Soloninupdated 25 September 2026

Mode

Text

Result

ResultSGVsbG8sIFVuaVRvb2xzIQ==
Result length24

Text to Base64 and back, without mangling non-Latin characters

Encode text to Base64 or decode it back, handling Cyrillic, accents and emoji correctly. Everything happens in your browser.

Base64 exists to push binary data through channels that only accept text — email attachments, data URLs, tokens embedded in JSON.

Naive implementations break the moment you feed them anything outside ASCII. This one encodes through UTF-8 first, so Cyrillic, accented letters and emoji survive the round trip intact.

How to use

Paste the text and pick a direction — encode or decode.
The result appears immediately, with its length beside it: that is the number to check against a field limit.
To verify a round trip, flip the direction and paste the result back; the original text should return character for character.

Why the growth is a third, and when it is not

Base64 counts bytes, not characters. Padded Base64 takes 4 × ceil(n ÷ 3) characters for n bytes: long inputs grow by about a third, short ones by more — 16 bytes become 24 characters, one byte becomes four. The rule holds for bytes; for text it depends on what the text is made of.

An accented letter already costs two bytes in UTF-8 before Base64 sees it, and an emoji costs four. That is why «naïve résumé» is twelve characters but fifteen bytes, and comes out as twenty.

Every row comes from the calculator on this page. The last one is the trap worth remembering: a single emoji is one character and four bytes, so it takes as much room as the word Hello.
Input	Characters	Bytes in UTF-8	Base64	Characters out
A	1	1	QQ==	4
ABC	3	3	QUJD	4
Hello	5	5	SGVsbG8=	8
café	4	5	Y2Fmw6k=	8
naïve résumé	12	15	bmHDr3ZlIHLDqXN1bcOp	20
🥨	1	4	8J+lqA==	8

The equals signs at the end are a counter

Base64 takes bytes three at a time and writes them as four characters. When the final group holds fewer than three bytes, the string is padded with equals signs — and the number of them tells you how many bytes were really there.

So the length of any padded Base64 string is a multiple of four, and no string ever ends in three equals signs: three bytes is already a full group.

Every row comes from the calculator on this page. Some libraries accept unpadded input, because the length alone says how many bytes are in the remainder — JSON Web Tokens drop the padding by rule.
Bytes in the final group	Equals signs	Example
3 — a full group	none	ABC → QUJD
2	one	AB → QUI=
1	two	A → QQ==

What this calculator shows

Encoded Base64, or the decoded original text
Correct UTF-8 handling for any alphabet
The resulting length, useful when a field has a size limit

What to keep in mind

Base64 is an encoding, not encryption — anyone can decode it instantly.
Output is roughly a third larger than the input, which is the cost of using only safe characters.

FAQs

Is Base64 a form of encryption?

No, and treating it as one is a real security mistake. Any Base64 string can be turned back with no key and no effort — which is exactly what this page does in the other direction.

Why does the encoded string end in = signs?

They are padding. Base64 works in groups of three bytes, and the equals signs mark a final group that was short: one sign means two bytes, two signs mean one.

Why does btoa fail on accented text?

Because btoa works on single bytes and throws on any character above code point 255: btoa("naïve €") fails on the euro sign, and btoa("café") returns Y2Fm6Q== — the Latin-1 byte of é, not UTF-8. The text has to be turned into UTF-8 first and only then encoded. This calculator does both steps, so a full round trip returns the original string.

Why do some Base64 strings use - and _ instead of + and /?

That is the URL-safe alphabet. Plus and slash already mean something in a web address — a slash separates path segments and a plus reads as a space in form data — so those two characters are swapped. This page uses the standard alphabet.

How much bigger does encoding make my data?

A third bigger, measured in bytes: three bytes become four characters. Short inputs look worse because of padding — a single byte occupies four characters — but the effect washes out with length.

Can I tell the original size from the Base64 length?

Almost exactly: bytes are about three quarters of the character count, minus the padding. Sixteen characters with no equals signs is twelve bytes; eight characters with two equals signs is four.

Worked example

A round trip through Cyrillic

Input: привет

Output: 0L/RgNC40LLQtdGC

Note: Decoding that returns the original word exactly. Implementations that skip the UTF-8 step return broken characters here.

Useful links

All tools in Dev Tools

Base64 Encoder / Decoder

Result

Related tools

URL Encoder / Decoder

Online Hash Generator

JWT Decoder

JSON formatter and validator

Useful links