Tests of different LLM models, local and API based.
Main focus is to find the best model and fine-tuning, to use for RoboCup @Home command generator.