What is Tool Calling?

A model is given a list of available tools, each described by a name, purpose, and a schema for its expected arguments. During generation, instead of producing a final answer directly, the model can emit a structured request to call one of these tools with specific argument values; the calling application executes the real function, returns the result to the model, and the model continues reasoning with that new information before answering the user. This loop can repeat across multiple tool calls within a single turn. Tool calling is what turns a language model from a text generator into an agent capable of taking real action, such as checking a calendar, sending an email, or querying a live database, grounded in current, verifiable data rather than only what the model memorized during training. Reliable tool calling depends on well-specified argument schemas and on the calling application validating and sandboxing what the model is allowed to actually execute.