convert numbers between any two bases from 2 to 4096, or encode text in any of them: base64, base32, hex, base1024 and beyond.
base 32 and base 64 use their standard RFC 4648 alphabets (A-Z 2-7, and A-Z a-z 0-9 + /), so results match other base32 and base64 tools. other bases count 0-9, then a-z, then A-Z, and case only matters past base 36. above 62 there is no standard, so digits continue into CJK characters starting at U+4E00 (一, the character for "one"), one character per digit up to base 4096.
text is encoded as UTF-8 bytes. power-of-two bases read those bytes as a stream of bits, a fixed number per character: 6 for base64, 9 for base512, 10 for base1024, 11 for base2048, 12 for base4096. other bases read the bytes as one big number and write a 0 for each leading zero byte, the way base58 does.
because RFC 4648 defines it for them and other tools expect it. no other base needs it. up to 8 bits per character the leftover is always under a byte, so the length is never ambiguous. above 8 bits, a short last chunk could look like an extra zero byte, so, like qntm's base2048, the last character can come from a small separate set that carries fewer bits: 2 characters for base512, 4 for base1024, 8 for base2048 and 16 for base4096.
yes, in number mode. above base 62 (except 64), separate the digit values with spaces, like 12 4000 7 in base 4096. the result is shown the same way under the output.
a leading minus sign works. fractions are not supported. numbers use arbitrary precision, so very long values convert exactly.