What Is Function Calling in AI?

Function calling (also called tool use) allows a large language model to invoke external functions, APIs, or tools as part of generating a response.

How It Works

1. You define available functions with their parameters (name, description, input schema)
2. The model decides when a function call would help answer the user’s question
3. The model outputs a structured function call (not free text)
4. Your application executes the function and returns the result
5. The model incorporates the result into its final response

Example

User: “What’s the weather in Tokyo?”

Instead of guessing, the model calls:

{
  "function": "get_weather",
  "arguments": { "city": "Tokyo" }
}

Your app runs the function, returns {"temp": 22, "condition": "sunny"}, and the model responds: “It’s currently 22C and sunny in Tokyo.”

Function Calling vs. MCP

Function calling is the mechanism — the model’s ability to output structured tool invocations. MCP is a protocol that standardizes how tools are defined and connected across different AI clients.

Supported Models

Model	Function Calling
GPT-6 Astra / GPT-5.6	Yes (native)
Claude Sonnet 5	Yes (tool use)
Gemini 3.8 Flash	Yes
Qwen3	Yes (with tooling)
Llama 3.2	Yes (with tooling)

Function Calling in Elvean

Elvean supports agentic tool calling across all compatible models — web search, image search, charts, maps, and any tools exposed via MCP servers.