Thank you very much for your answer. The uniqueness of PocketPal A is that the user does not need to look for models and spend hours checking their answers. The application has ready-made solutions. In other applications, the user has to iterate through the models, this is a very long process))
PocketPal AI seems like a good option for running LLMs locally on Android. I think RAM and processor make a big difference when choosing a model, especially on phones with 4GB or 8GB RAM. It would be helpful to compare a few lightweight models based on speed and response quality.