@FallenOne58035@QwenDevs yeaa but i have noticed significant drop in the text generation in the images like a word in the image it gets smudged but yea i totally get your point
@FallenOne58035@QwenDevs well 2512 does not work on 8gb i also have RTX 4060 but i feel that 16GB vram is the sweet spot i even have converted the model into API and give it for free for image gen and editing when i am not using my PC