Strategies to reduce the cost of running LLMs. Learn about batching, spot instances, and choosing the right model size for the job.