Run many prompts through one loaded Aspose.LLM model — amortize the load cost, pick session-per-prompt vs shared-session patterns....is the right pattern for classification, tagging, summarization...independent tasks like classification or per-item summarization...