I wanted to start a broader future possibilities thread around GPUs for Local AI.
A useful angle is future improvements, but it also connects with limitations. Different tools and workflows can produce very different results even when the goal is similar.
It would be interesting to compare...