We understand your frustration and appreciate your feedback. Our goal is to offer flexibility by allowing users to bring their own models to our platform and be able to use them on a shared pool of GPU machines, swapping them can certainly affect load times. This trade-off supports a wide range of creative possibilities. If we limited options to a single model, we could optimize for speed, but we prioritize versatility. In contrast, services like MimicPC may require additional time for model uploads, which can also impact overall workflow efficiency. We're continuously working to improve performance and appreciate your patience