After exiting MATLAB, it doesn't truly terminate in the Task Manager.

After exiting MATLAB, it doesn't truly terminate in the Task Manager.

Even after closing the MATLAB software (including using the exit command), the memory usage remains the same as when MATLAB was running, and the related processes don't exit.

When MATLAB is reopened, new related tasks are created.

Consequently, after repeatedly opening and closing the software several times, the memory becomes completely full, causing the computer to malfunction.

[Heran Wang]

[Heran Wang]

What on earth is going on? Why can't I completely shut down MATLAB?

[Matt J]

Is this Windows 11?

[dpb]

Are there any processes after a reboot, maybe?

There seem to be so many issues from R2024 on, I've not ventured past R2022b and am still on Win10 so no direct applicability to compare, but I wonder if Mathworks has tried to do something clever about having a "fast start" stub remain in memory that has issues.

I don't see any processes left after opening/closing either R2021b or R2022b in Win10 for comparison with earlier releases/OS(?) behavior.

[Heran Wang]

Windows 11

[dpb]

What about any processes loaded on starrtup?

I think your only chance here is to call Mathworks support directly. It would seem if this were a universal problem, it would have raised a large number of complaints before...

[Heran Wang]

I encountered the same issue on two computers running Windows 11: high memory usage that couldn't be turned off, which made it impossible to use them smoothly. I had no choice but to reluctantly roll back to the 2024 version.

[Stephen23]

Contact technical support. Make a bug report:

https://www.mathworks.com/support/contact_us.html

If you do not tell TMW what problems you experience with R2025a, then they cannot fix them.

[Heran Wang]

Is there anyone else with a similar situation? Or am I the only exception?

[dpb]

Certainly nobody else has chimed in with a "Me, too!" and as I observed previously, it would seem if this were a widespread phenomenon it would have been a major issue. As also noted, I think the only possible solution here is to contact TMW support; the Answers forum is not the official support channel.

[Alireza Esmailnezhad]

I recently installed Matlab 2025a, and I have the same trouble, not only that, it also consumes higher RAM than the previous releases. I think it is because of New Desktop (theme) of Matlab.

[dpb]

As suggested, above, submit this to Mathworks as an official support request/bug at <Product Support Page>

[Janis]

having same problem with command line calls with -r and -batch. Same in 2025a and 2025b. I'm gonna report this. Matlab.exe processes are left open. Matlab window might close (-r). With -batch it doesn't finish (powershell stuck)).

I realized in my CI pipelines. Workaround ist to let matlab taskkill itself (exit and quit don't work properly) and read exit code from a previously written .txt .

(Also described here MATLAB R2025a on Windows 11 [background open after closed] - MATLAB Answers - MATLAB Central)

[Walter Roberson]

I am not observing this on Windows 10.

[Walter Roberson]

I am not observing this on Windows 11 either. I do see that the MATLAB process takes several seconds to disappear, but not that many.

[Heran Wang]

MATLAB 2025b doesn't have this issue! A very good release.

[Janis]

I tested a little further. Correction: It has nothing to do directly with -batch -r command line callls. Happens also with open Matlab UI.

At least: Matlab 2025a and b on Windows Server 2025 generally has a problem quitting after having run our simulation.

I open Matlab UI. I run simulation. It finishes as it should. I execute quit(0, 'force') or without force -> UI closes but idling matlab.exe processes remain open and occupy RAM.

Never happened in years with simulation in 2019b.

[Walter Roberson]

Datapoint:

I installed Windows Server 2025 Standard Evaluation inside a virtual machine, and installed MATLAB R2025b inside of it.

There was a notable delay in cleaning up the MATLAB process after quitting, but the most old processes I ever got was 2 (with one of them promptly ending.) One of the processes did seem a little stuck at first, but went away when I waited longer.