Explore the world of open-source Large Language Models (LLMs) and their practical applications in this 54-minute conference talk by Andrey, founder of dstack. Gain insights into the comparative analysis between open-source LLMs and proprietary options like OpenAI, and delve into the economic aspects of hosting these models. Learn how to evaluate and benchmark LLM models based on performance, and witness a demonstration of operating and fine-tuning models using dstack, an open-source tool. Discover the costs and benefits associated with hosting open-source LLMs, and understand their potential for production environments. This talk, presented by LLMOps Space, a global community for LLM practitioners, provides valuable knowledge for those interested in deploying LLMs into production.