Note that the only requirement to run models is memory size. Theoretically you could buy the cheapest/slowest GPU out there, as long as it has sufficient memory for whatever models you want to use. Performance itself is about how long you’re ready to wait for 1 image, which can vary anywhere between few seconds and minutes.
I’ve heard people just buying Mac mini’s due to their memory size (unified), even tho the performance is not great, but you can run larger models as opposed to RTX 4060ti
Note that the only requirement to run models is memory size. Theoretically you could buy the cheapest/slowest GPU out there, as long as it has sufficient memory for whatever models you want to use. Performance itself is about how long you’re ready to wait for 1 image, which can vary anywhere between few seconds and minutes.
I’ve heard people just buying Mac mini’s due to their memory size (unified), even tho the performance is not great, but you can run larger models as opposed to RTX 4060ti