You mean, fine-tuning? If so, grounding LLMs is cheaper: https://techcommunity.microsoft.com/t5/fasttrack-for-azure/g...