RAMKeeper
How much RAM do I need on a Mac?
Nobody can answer this for you, but your current Mac can. It takes three commands and a week of normal use.
Every answer to this question is really an answer to a different one: how much RAM did the person replying need. That is why the advice ranges from "8 GB is plenty in 2026" to "never buy less than 32", and why both camps can produce screenshots to prove it.
Your own machine keeps the evidence. This page is the method; the two pages it links to are the decisions it leads to.
The three tests
Run these under your real workload, over about a week, at the end of working days rather than
after a restart. Nothing here installs anything or needs sudo.
Test 1: is your Mac swapping?
sysctl -n vm.swapusageSwap is disk pretending to be memory. Zero swap, or a small figure that never grows, means the memory you have is keeping up. Gigabytes every day means it is not. This single reading settles most cases.
Test 2: what does pressure say, not free memory?
memory_pressure -QFree memory is the figure people quote and the one that answers nothing: on the machine these guides are measured on, free memory was 0.12 GB out of 64 GB at a moment when swap was zero and the system reported 64% free by pressure. Both were correct. An operating system that leaves memory idle is wasting it.
Test 3: what do your apps peak at?
top -l 1 -o mem -n 10 -stats pid,command,mem
/usr/bin/footprint -p <pid>The second command prints the high-water mark for that process since it started, which Activity Monitor does not show anywhere. Current size tells you what an app costs now; the peak tells you what it will cost you again.
Turning a week of readings into a number
Take the apps you genuinely keep open at the same time, and add their peaks rather than their current sizes. From the worked reading, three of the measured peaks were:
| App | Current | Peak of its largest process |
|---|---|---|
| Vivaldi | 3.61 GB | 1.71 GB |
| Docker | 0.47 GB | 2.75 GB |
| node | 0.24 GB | 0.29 GB |
Two things in that small table are worth carrying with you. Docker's peak was six times its resting size, so any sizing based on what it looked like at that moment would have been wrong. And Vivaldi's peak is lower than its current total because the peak shown is for a single process while the current figure sums fifteen of them: a browser spreads its cost across renderers, so for browsers the group total is the number to use.
That gives you a working figure: the sum of what your usual set costs at their worst, plus room for macOS itself. That second part is not a rounding error and it is not a constant: on the machine measured here, wired memory, which applications cannot use at all, was 12.4 GB under a heavy load. It scales with what the system is doing, so treat it as a reading to take rather than a number to copy.
Where the answer points
| What your week looked like | Where to go next |
|---|---|
| You own an 8 GB Mac and want to know whether to keep living with it | Is 8 GB enough, including what actually helps if you already own one |
| You are buying and the choice in front of you is 16 against 24 GB | 16 GB vs 24 GB, with the thresholds that decide it |
| The numbers themselves confuse you: footprint, RSS, compressed, wired | What each figure in the Memory tab means |
| One app dominates everything else you measured | That is a different problem from being short of memory. See the current against peak section |
If you do not own the machine yet
Measure on whatever Mac you have now and read the result as a direction rather than a verdict. A machine with less memory than the one you are considering will overstate the problem, because it is already compressing and swapping to cope. A machine with more will understate it, because it has never had to. What travels between machines is the shape of your workload: which apps, how many at once, and what they peak at.
One hard constraint is worth repeating: on Apple silicon the memory is part of the chip package. There is no upgrade later, at any price, so this is a decision made once per machine.
What no tool can do for you
If the readings say you are short of memory, you are short of memory. No utility, ours included, creates RAM that is not there, and the category of app that offers to free some up with a button is measuring the wrong thing entirely. We took that apart separately.
What software is good for is the other case: when one application is responsible for most of your problem, every day, and you would rather set a rule once than watch for it. That is what RAMKeeper is, and its free build does the measuring half in the menu bar. The commands above give you the same numbers by hand.
Disclosure: published by RAMKeeper, which sells a Mac memory tool. Every command here ships with macOS, and the figures quoted were measured on 11 September 2026 on the machine described in the measured reading guide.