Decode a token, an e-mail header or a data URL, or make one. Non-Latin text is handled correctly.
It is a way of writing bytes as text that anyone can reverse. Never use it to hide passwords or personal data.
Many encode characters directly instead of their UTF-8 bytes, which breaks on non-Latin text. Here text is converted to UTF-8 first, so results match other languages and systems.
For URLs and JWTs, + and / become - and _, and trailing = is dropped. Decoding accepts both forms.
No. Pasting a token here keeps it on your machine.
5 MB for data URLs; Base64 is about 33% larger than the original.
The data was probably binary (an image or archive), not text.