"1.32 MB" Is that pronounced, "one-point-three-two" megabytes, or "one-point-thirty-two" megabytes?

Usually one point three two

I grew up with science classes telling us always state the digits individually. One point three two.
- Math class taught me to be precise I should always say "1 and 32 hundredths Megabytes"
  
  I don't think that's any more precise, just more verbose (read: inefficient).

One point three two. To me, thirty two is an integer.
- The only way you could use 'thirty two' correctly for that number would be 'one and thirty two hundredths' which would be pretty unusual.
- Agree. For things like semantic versioning, in which "1.20.1" and "1.2.1" are two different things, you want to pronounce them "one point twenty point one" and "one point two point one", respectively. But that is a bit of an outlier. File size should be pronounced "normally", because "1.20" and "1.2" are the same value.
  
  I disagree. I would personally find one point two zero point one to be more natural and easier to understand.

The first one is correct as others have said, but the second one is not ambiguous enough to confuse anyone nor weird enough for anyone to bat an eye at, you're fine with either.
- I'd say the second one is more correct, it sounds so weird pronouncing the digits singularly
  
  I’d say the second one is more correct
  In this case, it's not about what sounds good or personal opinion, there is a standard name for that number for a reason. If I go around calling 100 "one oh oh" or "tenty ten", it's clear what number I mean but I can't honestly call it more correct, because there's a standard English name for it.
  To demonstrate a part of why it's clearer that way, put these numbers in ascending numerical order: (e.g. 1, 2, 3, ... )
  one point three
  one point twenty-nine
  one point thirty
  one point thirty-one
  one point three-thousand-and-fifty-two
  Hopefully this clarifies that we're not actually dealing with a "thirty-two" when we're talking about 1.32 (edit: that said, when we're talking about version numbers, e.g. Linux kernel 4.20, which is greater than Linux kernel 4.9, then we'd say "four point twenty")

Ten-million-five-hundred-and-sixty-thousand bits.

One point four four
- I have heard people drop the "point" and say "One Fourty-four".
  
  Tree fiddy!
  
  criminal!
  “One fourty-four Em Bee floppy”

I mostly heard it one point thirty two? Grew up in Sweden, living in France. If someone says one point three two I'd assume they're Americans.
I might be totally wrong, just stating what I have heard
- No that's interesting, I was wondering if there was a cultural divide.
  Thirty two sounds so alien to me, but I heard it in a Nerdstalgic video and wondered if it was an American thing
  
  Definitely, in frech itd be un point trente-deux mégaoctets or 1.32mo
  edit: forgot not everyone speaks french, the french version is one point thirty-two
- I had the same experience (also European), but didn't know the Americans changed it specifically for bytes
  
  We don't. That's just the normal way most people pronounce numbers with a decimal point. The big exception is prices: $1.32 is often pronounced "one thirty two".

Decimals are usually spelt out a digit at a time. 3.14159 would be three point one four one five nine, not three point fourteen thousand one hundred and fifty nine. 37.32 would be thirty-seven point three two. If it's not a decimal but something like a version string then you could say v3.14 is version three point fourteen, and three point one four might be confused with 3.1.4 even though you didn't say the second point. IP addresses are a bit mixed; I'd say ten ten, but also one nine two dot one six eight.

The former.

One point three two, or one three two if it's obvious from context where the decimal point is. That's how you're meant to pronounce digits after the decimal point in general.

I agree that the precision is not that valuable as some have said. I'd just read the numbers off as one point two three megabytes since anyone who cares can reconstruct the number, anyone who doesn't can stick to the first few sig figs.
For 257.62 GB I'd say "two hundred fifty seven point six two". Yep. I put in the effort for the most significant of the digits, I dont bother beyond that.
8249.19 GB? About 8 terabytes. Doesnt really matter anymore.

"about a meg" because it's almost unthinkable anyone cares about 3 tenths of a meg much less 2 hundredths.
- "about a meg" because it's almost unthinkable anyone cares about 3 tenths of a meg much less 2 hundredths.
  Tell me you never used floppy discs as a storage medium without telling me.
- I'd round up to one and a half. Also remove "bytes" and "bites". 1.32 MB is "one and a half megs" or even "a meg and a half"

The first one

Canada (Ontario) here. Was taught explicitly to say "point three two"

One and thirty-two hundreths

Its pronounced 'About four thirds megabytes.'

growing up with floppy disks and diskettes on the east coast US it was 'one point two megabyte' and 'one point four four megabyte' exclusively

"One point three two", because otherwise the question is 'thirty two what'. Consider what happens if we put a zero on the end — does it become "one point three hundred and twenty" despite being exactly the same number?
- Yeah and what is 1.032?
  
  A rounding error
  I'd probably say "one point oh three two" for that one though

First question, and it’s important: Are you Doc Brown?

Neither.
It's pronounced: "one and thirty-two hundreths of a megabyte". Properly.
But idgaf how you pronounce it as long as I understand exactly what you're saying. Personally, "one point three two".

About one floppy disk, with a little free space to spare.

Very little, around 60k.

A 1.44 "MB" floppy is 1440k, or about 1.406 real MB, and of that the space used by the FAT file system reduces it to around 1.38 free space.
For some reason I couldn't find the exact number and don't have any handy to check it myself.

The floppy disk format is based on the FAT12 file system.
https://www.cs.drexel.edu/~johnsojr/2012-13/fall/cs370/resources/UnderstandingFAT12.pdf
And with enough creative tweaks to that file system, you can get DMF 1.68MB format, and if you think a bit outside the box and erase the redundant secondary FAT table and settle on a max of only 16 files on the disk, you can squeeze a few more kilobytes out of that even.
I actually made a number of custom modded blank disk images with more storage space, I might dig out the full specs of all the variants later.

    

Modified versions of various blank floppies
-------------------------------------------
These modifications reduce the number of
FAT tables from 2 to 1 and also reduce the
number of root entries down to 16 files,
which frees up some extra storage space.

The 1.72MB format can ONLY be used on Win9X
systems on real hardware, as not even WinNT
can access tracks 81 or 82 on floppy disks.
Disk image programs like WinImage can still
access files within 1.72MB floppy images.



1.44MB Standard:
80 Tracks		18 Sectors/Track
2880 Sectors Total	1474560 Bytes Total
-------------------------------------------
Sectors Per Cluster: 1	Number of FATs: 2
Max Root Entries: 224	Sectors Per FAT: 9
1457664 Bytes Data

1.44MB Maxed:
-------------------------------------------
Sectors Per Cluster: 4	Number of FATs: 1
Max Root Entries: 64	Sectors Per FAT: 3
1470464 Bytes Data

Differences:
-------------------------------------------
12800 Bytes More, 160 Less Root Entries



1.68MB Standard:
80 Tracks		21 Sectors/Track
3360 Sectors Total	1720320 Bytes Total
-------------------------------------------
Sectors Per Cluster: 1	Number of FATs: 2
Max Root Entries: 224	Sectors Per FAT: 10
1702400 Bytes Data

1.68MB Maxed:
-------------------------------------------
Sectors Per Cluster: 4	Number of FATs: 1
Max Root Entries: 64	Sectors Per FAT: 3
1716224 Bytes Data

Differences:
-------------------------------------------
13824 Bytes More, 160 Less Root Entries



DMF 1024 Standard:
80 Tracks		21 Sectors/Track
3360 Sectors Total	1720320 Bytes Total
-------------------------------------------
Sectors Per Cluster: 2	Number of FATs: 2
Max Root Entries: 16	Sectors Per FAT: 5
1714176 Bytes Data

DMF 1024 Maxed:
-------------------------------------------
Sectors Per Cluster: 4	Number of FATs: 1
Max Root Entries: 64	Sectors Per FAT: 3
1716224 Bytes Data

2048 Bytes More, 48 More Root Entries



DMF 2048 Standard:
80 Tracks		21 Sectors/Track
3360 Sectors Total	1720320 Bytes Total
-------------------------------------------
Sectors Per Cluster: 4	Number of FATs: 2
Max Root Entries: 16	Sectors Per FAT: 3
1716224 Bytes Data

DMF 2048 Maxed:
-------------------------------------------
Sectors Per Cluster: 4	Number of FATs: 1
Max Root Entries: 64	Sectors Per FAT: 3
1716224 Bytes Data

Differences:
-------------------------------------------
0 Bytes More, 48 More Root Entries



1.72MB Standard:
82 Tracks		21 Sectors/Track
3444 Sectors Total	1763328 Bytes Total
-------------------------------------------
Sectors Per Cluster: 1	Number of FATs: 2
Max Root Entries: 224	Sectors Per FAT: 10
1745408 Bytes Data

1.72MB Maxed:
-------------------------------------------
Sectors Per Cluster: 4	Number of FATs: 1
Max Root Entries: 64	Sectors Per FAT: 3
1759232 Bytes Data

Differences:
-------------------------------------------
13824 Bytes More, 160 Less Root Entries

If you're interested in the blank disk images themselves, let me know.

Also, 1474560 / 1024 = 1440
If anyone could keep up with binary numbers back in the day, floppy disks were literally measured in binary megabytes.

Jeden przecinek trzydziesci dwa

Only time I can think of where the 32 of 1.32 could be said as thirty-two would be as a software version number

Depending on the necessary precision it could be "a meg and change" 😁

One and thirty two hundredths of a megabyte
Or
One megabyte three hundred and twenty kilobytes

One point three two emm bee 😁

I'd say one point thirty-two. As others noted, much depends on geography.
Personally, I say the "actual" number up to 3 or 4 decimal places, with a lot of the reason depending on the specific context. If I had to asses, I'd say I say the "whole" number in over 50% of cases for 3 digits, and in about 10% for 4 digits. Anything over 4 decimal places and I fall back to individual digits.