How to make `ffmpeg` use GPU mostly for reducing file sizes?

I'm on Debian 12 & I'm using a ryzen 5600H with integrated graphics + Nvidia 1650 setup & I'm running the following command

ffmpeg -y -hwaccel cuda -i input.mp4 -vcodec libx264 -crf 24 output.mp4

But this mostly uses CPU, as I can see all of the 12 threads are getting to 100% and bloating other tasks. How can I make ffmpeg use my NVIDIA GPU mostly for this task? I have searched through various posts before, but they all seemed to be very outdated, and the ones I was able to make work destroyed the quality.