Tool use / function calling
Function calling is the mechanism. Tool use is the capability it enables. The model never executes anything itself.
The two terms are interchangeable and describe the mechanism that allows an LLM to take action outside the confines of its training data. The model itself does not execute code or call APIs. It generates a structured request which the harness executes, returning the results back to the model. From a governance perspective the terms carry a slightly different emphasis. Function calling refers to the constrained interface where only defined functions with specified inputs can be invoked. Tool use is the broader concept describing the full set of external capabilities an agent can utilize. Function calling is the mechanism. Tool use is the capability it enables.