Which statement correctly describes function calling in LLMs?

• A. It allows the model to call external predefined functions
• B. It updates the training data
• C. It changes the network architecture
• D. It stores logs of calls

Answer: (A)

Function calling is a feature that lets an LLM interact with external tools or services. At runtime, the model can output a request specifying a function name and arguments. The system then executes that predefined function (such as a calculator, a weather API, or a database query) and returns the result to the model to continue the conversation. This expands what the model can do beyond text generation by fetching fresh data or performing actions it can’t do on its own. It doesn’t change the model’s training data or its underlying architecture, and logging of calls is handled by the surrounding system rather than being a defining part of function calling itself. So the statement that describes function calling as allowing the model to call external predefined functions is the correct description.