Number Base Converter.
A different base. The same value.

Choose an input base from 2 to 36. Digits are 0–9 and A–Z, case insensitive, with an optional leading plus or minus. The matching 0b, 0o or 0x prefix is accepted for bases 2, 8 or 16. A recognized prefix for a different base is rejected, even if its letters are valid digits: enter hexadecimal 0b10 as 0x0b10. Whitespace around each value is trimmed. Fractions, separators and internal whitespace are rejected. There is no scientific-notation interpretation: 1e3 is invalid decimal but is a valid hexadecimal integer.

Integer arithmetic uses BigInt and exports numbers as strings. Conversion accepts up to 1,024 digits per value, excluding the sign and matching prefix. It does not round through floating-point arithmetic. Negative conversion outputs use a minus sign; two’s-complement patterns are shown separately only in bitwise mode. These are integer conversions, not floating-point encodings.

Bitwise mode supports 8, 16, 32, 64, 128 or 256 bits. Unsigned input must be from 0 to 2^width − 1; signed input must be from −2^(width−1) to 2^(width−1) − 1. Out-of-range input is rejected instead of silently truncated. Operand B uses the same input base and signedness as operand A. Results show the selected signed or unsigned interpretation, both decimal interpretations, a padded binary pattern, padded hexadecimal and bytes in both byte orders.

AND, OR, XOR and NOT operate on the chosen fixed-width pattern. Left shift discards bits beyond the width; logical right shift inserts zero bits. Arithmetic right shift repeats the pattern’s highest bit, regardless of the input signedness setting. Rotations wrap bits around within the chosen width. Shift counts are decimal integers from 0 through the width. Shifting by the full width yields zero, except arithmetic right shift yields all sign bits; rotating by the full width leaves the pattern unchanged.

Base conversion accepts up to 1,000 nonempty lines and 256 KiB of UTF-8 input. Blank lines are skipped and original line numbers are retained. UTF-8 file import omits a leading BOM and rejects malformed encoding. Results show 50 rows per page; complete exports contain every row. Bitwise mode calculates one operation at a time. Local workers have 5-second startup and 10-second processing deadlines. Cancel discards partial results.

Copy valid results or download TXT in the selected output base. Copy a single valid row, an inspected row as JSON, or the entire JSON report. JSON and CSV retain invalid-row diagnostics; JSON also retains original input, while CSV omits that untrusted column. Output base and letter-case changes update exports without rerunning the calculation. Inspector selection and pagination keep download links unchanged. CSV adds an apostrophe before negative outputs for spreadsheet safety; JSON and TXT keep raw values. Import all result columns as text in spreadsheet software to preserve exact digits, letter-based values and leading zeros.

No. Big-endian bytes place the most significant byte first; little-endian bytes place it last. The inspector shows both sequences for the same fixed-width bit pattern. Ordinary base conversion has no assumed bit width, padding or byte order. Choose bitwise mode when you need a defined two’s-complement representation.

Calculations and file reading happen locally in your browser. This tool does not upload input or save history. Clear removes input, results and temporary download links, and leaving the page clears tool state. Normal site analytics and feedback services may still load. A modern browser with BigInt and Web Workers is required.