AI & LLM

LLM Tool / Function Call JSON Example

A JSON example of LLM function calling — the tool definition you send and the tool_call the model returns with arguments. Copy-ready for OpenAI and Anthropic tool use.

JSON

Generate mock data Open in Formatter

{
  "id": "call_9aZ2kQ7rT1vBnM3p",
  "type": "function",
  "function": {
    "name": "get_weather",
    "arguments": "{\"location\":\"Ahmedabad, IN\",\"unit\":\"celsius\"}"
  }
}

Field Reference

idrequiredstringUnique id for this tool call; echo it back in the tool result message

typerequiredstringCall type — 'function' for function calling

function.namerequiredstringName of the tool the model chose to call; must match a defined tool

function.argumentsrequiredstring (JSON)Arguments as a JSON-encoded string — parse it before use; the model may hallucinate fields

Variants

Tool definition (sent to the model)The JSON Schema you register so the model knows the tool's parameters.

Tool definition (sent to the model)

Generate mock data Open in Formatter

{
  "type": "function",
  "function": {
    "name": "get_weather",
    "description": "Get the current weather for a city.",
    "parameters": {
      "type": "object",
      "properties": {
        "location": {
          "type": "string",
          "description": "City and country, e.g. 'Ahmedabad, IN'"
        },
        "unit": {
          "type": "string",
          "enum": [
            "celsius",
            "fahrenheit"
          ],
          "default": "celsius"
        }
      },
      "required": [
        "location"
      ]
    }
  }
}

Tool result (sent back to the model)After running the function, return its output referencing the call id.

Tool result (sent back to the model)

Generate mock data Open in Formatter

{
  "role": "tool",
  "tool_call_id": "call_9aZ2kQ7rT1vBnM3p",
  "content": "{\"temperature\":34,\"unit\":\"celsius\",\"conditions\":\"Sunny\"}"
}

Common Use Cases

→Letting an LLM trigger real functions like database lookups or API calls
→Building agents that decide which tool to call and with what arguments
→Forcing structured output by defining a single tool the model must call

function callingtool usellmopenai toolsagentsjson schema

Validate or format this JSON

One click loads this exact example into the tool — no copy-paste needed. Format it, validate it, explore the tree, or generate TypeScript types instantly.

Format Validate Explore tree Generate TypeScript Minify

Frequently Asked Questions

The model generates arguments as text, so APIs return them as a JSON-encoded string. Always JSON.parse() the arguments and validate them against your schema before executing — the model can omit required fields or invent values.

Define a single tool whose parameters describe your desired shape and set tool_choice to force that function. The model then must respond with a tool_call whose arguments match your schema, giving you reliable structured output.

They're the same concept with different shapes. OpenAI returns a tool_calls array on the message with function.name and a stringified function.arguments; Anthropic returns a tool_use content block with name and a parsed input object. The orchestration loop — call tool, return result, continue — is identical.

Related JSON Examples

AI & LLMOpenAI Chat Completion ResponseOpenAI Chat Completions API response with choices, message, finish_reason, and usage.AI & LLMAnthropic Claude Messages ResponseClaude Messages API response with content blocks, stop_reason, and token usage.AI & LLMAI Agent Run TraceAgent run with goal, reasoning steps, tool calls, observations, and final output.