Imagine trying to find a needle in a haystack. That's often what it feels like for engineers verifying complex systems. They have to meticulously examine massive models to ensure they meet specific requirements, a process that's both time-consuming and prone to errors. But what if there was a way to automatically isolate the relevant parts of a model, making the verification process significantly faster and more efficient? That's the promise of a new technique using Large Language Models (LLMs) to 'slice' Simulink models. Simulink, a popular tool for modeling complex systems, uses visual block diagrams to represent the system's behavior. This new research explores how LLMs can understand these diagrams, converted into text, and pinpoint the specific blocks needed to satisfy a given requirement. Think of it like asking the LLM, 'Which parts of this model are responsible for ensuring the brakes work correctly?' The LLM then analyzes the textual representation and returns a 'slice' of the model containing only the essential blocks. This research dives into different ways of converting the visual model into text, finding that a 'medium-verbosity' approach works best. This means keeping the essential information about the blocks while discarding unnecessary visual details. The study also experimented with different prompting strategies, discovering that 'chain-of-thought' and 'zero-shot' prompting are most effective for guiding the LLM to accurate slices. This approach has the potential to revolutionize how engineers verify complex systems, saving time and reducing errors. However, there are still challenges to overcome. The research highlights the need for further testing with more diverse Simulink models and different LLMs. It also emphasizes the importance of carefully evaluating the accuracy of the generated slices. As LLMs continue to evolve, their ability to understand and analyze complex systems like Simulink models will only improve, paving the way for more efficient and reliable system verification in the future.