Part 3 Advanced topics: Deeper understanding and customization
By now you have a local AI setup that runs and answers your prompts. Part 3 is about the decisions you make once the basic setup already works: which model to use, how to shape its behavior, and how to trust what it gives you.
Chapter 12 compares models side by side, not only through benchmark scores but also through the practical tradeoffs that matter to you: speed, memory use, and answer quality. Chapter 13 shows you how to control that behavior directly, through system prompts, parameters like temperature and context length, and Modelfiles that let you save a configuration as your own named model.
Running a model locally is not automatically safe or automatically correct, and chapter 14 deals with that directly. You will learn how to verify that a model actually works offline, what privacy guarantees local inference gives you and what it does not, basic security habits, and you’ll get a realistic look at the cost of running models on your own hardware versus paying for an API. Chapter 15 is a troubleshooting chapter focusing on problems that actually happen: models that will not load, answers that come out garbled, and hardware that runs out of memory partway through a session.