How this calculator works
Put a number in X, a number in Y where the operation needs one, and press the operation you want. The answer appears in full, however long it turns out to be — no E-notation, no rounding, no digits quietly dropped off the end.
Underneath, nothing is stored as a decimal fraction at all. Every value is held as a whole number together with a note of where the decimal point sits, so 12.75 is kept as the pair 1275 and two. Addition lines the points up, multiplication adds the two positions together, and the arithmetic itself runs on integers that can grow to any length your device will hold. That is the whole trick, and it is why the hundredth digit of an answer here is as trustworthy as the first.
12.75 = 1275 ÷ 10²
a × b = (na × nb) ÷ 10placesa + placesb
Where ordinary calculators give up
A phone, a spreadsheet and most programming languages all store numbers the same way: a fixed budget of about 15 to 17 significant digits, with everything past that rounded away. For daily arithmetic this is invisible. Push slightly beyond it and the answers stop being right.
| Calculation | Typical calculator | Exact answer |
|---|---|---|
| 123456789123456789 × 2 | 246913578246913570 | 246913578246913578 |
| 9007199254740993 + 1 | 9007199254740992 | 9007199254740994 |
| 25! | 1.551121004×1025 | 15511210043330985984000000 |
| 0.1 + 0.2 | 0.30000000000000004 | 0.3 |
The last row is the one that surprises people most, and it has nothing to do with size. A computer stores numbers in binary, and one tenth has no exact binary form for the same reason one third has no exact decimal form. The stored value is a hair away from 0.1 before any arithmetic starts, and the error surfaces as soon as you add. Because this page keeps 0.1 as the whole number 1 with one decimal place, that error never arises.
The eleven operations
Each button acts on the values currently in X and Y. The five on the right use X alone, so Y can be left empty.
| Button | What it does | Needs |
|---|---|---|
| X + Y, X – Y, X × Y | Exact to the last digit, whatever the size | Any numbers |
| X / Y | Division to the precision you set | Y not zero |
| X^Y | X raised to a whole power, negative allowed | Y a whole number |
| √X | Square root to the precision you set | X not negative |
| X ^ 2 | X multiplied by itself, always exact | Any number |
| X! | Every whole number from 1 to X multiplied together | Whole number, up to 20,000 |
| MOD | What is left over after dividing X by Y | Positive whole numbers |
| GCD | Largest number dividing both | Positive whole numbers |
| LCM | Smallest number both divide into | Positive whole numbers |
Factorials grow more violently than anything else on that list. 20! already runs past 18 digits, 70! past 100, and 1000! fills more than 2,500 digits. If a press seems to hang for a second on a large factorial, that is your own processor doing honest work rather than the page failing. For the divisor-and-multiple pair on their own, our GCF calculator and LCM calculator show the full working rather than just the answer.
Precision, and why digits are cut rather than rounded
Two of the eleven operations can run forever. One third is 0.333... with no end, and the square root of two never repeats or terminates. For those, the precision box decides where to stop. Every other operation is exact and ignores the setting entirely.
Where the answer stops, the remaining digits are cut off rather than rounded. Two thirds to five places gives 0.66666, not 0.66667:
2 ÷ 3 at 20 places = 0.66666666666666666666
√3 at 7 places = 1.7320508
Cutting is the safer default for a tool like this, because every digit shown is then a digit the answer genuinely has. A rounded final digit is the one place a result could disagree with a longer calculation, and the way to avoid the question entirely is to ask for a few more places than you need and read the ones you want.
Trailing zeros are dropped, so a division that comes out exactly keeps its natural length: ten divided by four is shown as 2.5 rather than 2.5 followed by nineteen zeros.
Entering very large and very small numbers
Typing out a twenty-digit number is tedious and easy to get wrong, so E-notation is accepted in either box. The letter E or e means "times ten to the power of", and what follows is the power.
| You type | It means | In full |
|---|---|---|
| 23E18 | 23 × 1018 | 23,000,000,000,000,000,000 |
| 3.5e19 | 3.5 × 1019 | 35,000,000,000,000,000,000 |
| 1e-5 | 1 ÷ 105 | 0.00001 |
| 6.022e23 | 6.022 × 1023 | 602,200,000,000,000,000,000,000 |
Commas in a pasted number are ignored, so a figure copied from a spreadsheet or an article can go straight in. Results always come back written out in full, which is the main reason to reach for a page like this instead of the calculator already on your desk.
What exact arithmetic actually means
"Exact" here has a precise meaning worth separating from "lots of digits". A result is exact when no step in producing it threw information away. Addition, subtraction, multiplication, squaring, factorials, MOD, GCD and LCM on whole numbers are all exact in this sense: the answer is the answer, not a close approximation to it.
Division and square roots are different in kind. The true answer often has no finite decimal form at all, so some cut is unavoidable and the only honest question is where. That is what the precision box is for, and it is why those two operations are the only ones it touches.
The practical consequence: if a calculation matters, prefer a route through the exact operations. Working out a ratio by dividing early and multiplying the rounded result compounds the cut; keeping the whole numbers intact and dividing once at the end does not.
Where big numbers genuinely matter
- Cryptography. RSA keys are built from primes hundreds of digits long, and every operation on them has to be exact, since a single wrong digit makes a key useless.
- Combinatorics. Counting arrangements produces factorials fast. A 52-card deck can be ordered in 52! ways, a 68-digit number.
- Astronomy and physics. Distances in metres, particle counts and Avogadro-scale quantities all run past what a normal calculator holds.
- Finance over long horizons. Compounding across decades, or in a currency with very small units, accumulates rounding error that exact arithmetic avoids.
- Number theory and puzzles. Mersenne primes, perfect numbers and digit-sum problems all need the full value rather than its first fifteen digits.
Edge cases and common mistakes
- Leaving Y filled on a one-number operation. √X, X ^ 2 and X! read X only. Whatever is in Y is ignored rather than causing an error.
- Expecting a fractional exponent. X^Y wants a whole exponent. For a square root, use the √X button.
- Setting precision to zero. The box needs a whole number of at least 1. If you want a whole-number answer from a division, MOD gives the remainder and the quotient is the division read before the point.
- Negative values in MOD, GCD or LCM. All three are defined on positive whole numbers here, so a negative input is refused rather than guessed at.
- Very large factorials. Anything past a few thousand produces a result with more digits than fits comfortably on screen, and takes real time to compute.
What this page does not do
It performs one operation at a time on two values. There is no expression parser, so a compound calculation has to be done in steps, carrying the result forward with the reuse button under the answer. It does not handle fractional exponents, logarithms, trigonometry or complex numbers, and it does not factorise a number into primes — our factor calculator covers that, and the scientific notation calculator converts between written-out and E-notation forms.
Everything runs on your own device, so very large factorials and ten-thousand-digit divisions are bounded by your processor and memory rather than by any limit set here.
Frequently asked questions
Why does my phone calculator give a different answer?
Most calculators hold about 15 to 17 significant digits and silently round everything beyond that. Ask one for 123456789123456789 × 2 and you will often get 246913578246913570, with the last two digits wrong. This page works in exact whole-number arithmetic, so no digit is ever approximated.
How many digits can it handle?
There is no fixed ceiling on the size of X or Y — they are limited only by your device's memory. Decimal places in a division or square root go up to 10,000. Factorials are capped at 20,000, which already produces a result over 77,000 digits long.
What does the precision box actually do?
It sets how many digits appear after the decimal point for the two operations that can run forever: division and square root. Everything else is exact and ignores it. The digits are cut off rather than rounded, so 2 ÷ 3 at five places gives 0.66666, not 0.66667.
Can I enter scientific notation?
Yes. Both 23E18 and 3.5e19 are understood, as are plain decimals, negatives and numbers with commas in them. Results are shown in full rather than in E-notation, which is the point of a calculator like this.
Why are factorial, MOD, GCD and LCM fussy about the input?
All four are defined only on positive whole numbers. There is no sensible factorial of 3.5 and no greatest common divisor of two decimals, so the page asks for a whole number rather than quietly rounding yours and returning an answer to a question you did not ask.
Is X^Y limited to whole exponents?
Yes. Whole exponents, positive or negative, are handled exactly: 2^-3 returns 0.125. A fractional exponent such as 2^0.5 is a root rather than a power, and asking for it exactly would need an infinite decimal — use the √X button for square roots.
Does anything I type get uploaded?
No. Every calculation runs in your browser using built-in big-integer arithmetic. Nothing is sent to a server, which is also why a 20,000-digit factorial takes a moment: your own device is doing the work.
This calculator is provided for general technical and educational use. Results are computed in exact integer arithmetic and verified against an independent reference implementation, but any figure that carries real consequences in engineering, finance or cryptography deserves a second check against a second tool. See our about page for how these tools are built and tested, and our privacy policy for what we do and do not collect.