GPT-4 is known for introducing which capability in LLMs?

• A. Image generation
• B. Rule-based reasoning
• C. Speech synthesis
• D. Function calling

Answer: (D)

GPT-4 introduced the ability for an LLM to call external functions, letting the model invoke tools or APIs during a conversation and then use the results in its next messages. This turns the model from just generating text into acting as a controller that can fetch data, perform computations, or trigger actions in other systems, then incorporate those results into the reply. For example, it can call a weather service to get current conditions, use a calculator tool for precise math, or query an API and present the combined information.

Image generation, speech synthesis, and purely rule-based reasoning describe other capabilities or approaches that aren’t the specific feature highlighted here. Image generation comes from image-producing models, speech synthesis is separate text-to-speech technology, and while reasoning methods matter, the notable addition GPT-4 brought is tool use through function calls.