/en/tool/base64/ works Encoded in the browser
Paste a string, drag a file onto the panel, or paste an image from the clipboard — the tool converts data to and from Base64. You can save decoded data as a file: the type is detected from the first bytes, and images and PDFs are displayed right away. It understands JWT, email headers such as =?windows-1251?B?…?=, and supports hex, Base32, and percent encoding in addition to Base64. Nothing is sent anywhere: everything runs in the browser.
A file can be dragged anywhere on the panel, and an image from a screenshot tool can be pasted with Ctrl+V.
Appears automatically when there is something to show: an image, PDF, audio, token, or just bytes.
Preview
JWT parsing
Signature not verified. The page only splits the token into three parts and reads them: you cannot verify the signature without a secret or public key, and you should not do this in the browser. Most importantly, a JWT payload is not encrypted — anyone who gets the token can read it. Do not put passwords, keys, or card numbers in the token.
Ready-to-use snippet
Hex dump
Choose a mode. “Encode” converts text or a file to Base64; “Decode” converts it back. You do not have to find the file in Explorer: you can drop it onto the panel, or paste an image from a screenshot tool with Ctrl+V.
The “Convert to” row switches the result alphabet without changing the input: Base64, Base64 for URLs, hex, Base32, and percent encoding are all calculated from the same bytes. During decoding, the same row specifies what to decode.
The decoded result is not necessarily text. If the first bytes identify an image, PDF, or audio file, the tool names the format, displays it right on the page, and offers to save it with the correct extension.
If the bytes do not form valid UTF-8, the tool tries windows-1251, KOI8-R, and CP866 and suggests the one that produces readable Cyrillic text. You can switch to it with one click.
You do not need to decode a JWT string or an email header such as =?windows-1251?B?…= separately: the tool recognizes them automatically and shows their parts.
The “String check” block works before any decoding: length, divisibility by four, alphabet, padding, and the resulting byte count. It shows whether the string is truncated or complete.
Base64 characters = 4 × ⌈Bytes ÷ 3⌉
That is why the increase is always one-third: three bytes of data become four characters. hex doubles the size exactly; Base32 increases it by half; the increase from percent encoding depends on the data: Latin letters and digits stay as they are, while everything else triples in size. Base64 is not encryption or compression, but a way to carry binary data through a text-based channel.
The input data is the same as the default field values: you can calculate it mentally and confirm that the tool is telling the truth.
Each of the six Cyrillic letters takes two bytes in UTF-8. That makes 12 bytes in total.
12 is evenly divisible by 3, so there are exactly four groups and no equals signs are needed at the end.
Output characters: 4 × (12 ÷ 3) = 16.
Increase: (16 − 12) ÷ 12 = 33%.
The result is 0J\/RgNC40LLQtdGC — 16 characters versus 12 bytes. With the URL-safe alphabet, the slash is replaced with an underscore: 0J_RgNC40LLQtdGC. If you switch the encoding to windows-1251, the same six letters take 6 bytes, and the string is half as long — 8 characters.
Paste a string, turn on “Decode,” and click “Download as file.” The type is detected from the first bytes — PNG, JPEG, GIF, WebP, PDF, ZIP, MP3, MP4, font, SVG, JSON — and the extension and media type are filled in automatically. Images, PDFs, and audio are shown right on the page before saving.
Drop an image onto the panel in encoding mode. A ready-made data:image\/…;base64,… string will appear below, along with three snippets, each with its own copy button: an img tag, a background-image property, and an @font-face rule for fonts. Keep in mind that a file in a data: URI is one-third longer and is cached with the stylesheet rather than separately.
No. The conversion is reversible without a key, and anyone can decode the string, including this page. Hiding passwords and tokens in Base64 is pointless: it is encoding, not protection. For the same reason, the contents of a JWT are plain text to anyone who has the token.
The bytes are not encoded in UTF-8. This is how exports from 1C, old emails, and data from systems written before 2005 often arrive. The tool detects this automatically: it tries windows-1251, KOI8-R, and CP866 and suggests the one that produces readable text. You can accept it with one click.
This is how RFC 2047 encodes non-ASCII text in email headers: the subject, sender name, and attachment name. Paste the entire string, including the word Subject. The tool will parse all the parts, both B and Q, using the appropriate encoding for each, join them, and show the finished header.
50 MB. Anything larger than 1.5 MB is processed in 3 MB chunks in a separate stream with a progress bar, so the tab remains responsive. The window shows the first 100,000 characters of the result: there is no point in rendering tens of millions of characters in a browser, while the “Copy” and “Download” buttons return the entire string.
Most often, extra characters were included during copying — quotation marks, non-breaking spaces, or a truncated ending. The tool shows exactly which character is not in the alphabet and where it appears. The “String check” block separately reports whether the length is divisible by four and whether the padding is valid.
Exactly. It is read by the browser directly on the page, just as any upload form does before you click “Submit.” The page makes no network requests — you can see this in the developer tools. Only the settings and short text, if it fits, appear in the address bar; files and long strings are never written there.
Creates passwords of the required length and assesses how difficult they are to crack.
Open