Skip to main content

​
Function Calling

Function Calling allows models to call functions you define during a conversation. The model does not execute functions directly; instead, it returns the function name and arguments, and your code executes the function and returns the result to the model.

​
Workflow

  1. Send messages and tool definitions to the model
  2. The model determines whether to call a tool and returns tool_calls
  3. Your code executes the function and gets the result
  4. Send the function result back to the model as a tool role message
  5. The model generates the final response based on the function result

​
Define Tools

Define available functions in the tools parameter: 
curl https://crazyrouter.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "gpt-4o",
    "messages": [
      {"role": "user", "content": "What is the weather like in Beijing today?"}
    ],
    "tools": [
      {
        "type": "function",
        "function": {
          "name": "get_weather",
          "description": "Get the current weather information for a specified city",
          "parameters": {
            "type": "object",
            "properties": {
              "city": {
                "type": "string",
                "description": "City name, e.g.: Beijing, Shanghai"
              },
              "unit": {
                "type": "string",
                "enum": ["celsius", "fahrenheit"],
                "description": "Temperature unit"
              }
            },
            "required": ["city"]
          }
        }
      }
    ],
    "tool_choice": "auto"
  }'

​
Model Returns tool_calls

When the model decides to call a function, the response looks like this: 
{
  "id": "chatcmpl-abc123",
  "object": "chat.completion",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": null,
        "tool_calls": [
          {
            "id": "call_abc123",
            "type": "function",
            "function": {
              "name": "get_weather",
              "arguments": "{\"city\": \"Beijing\", \"unit\": \"celsius\"}"
            }
          }
        ]
      },
      "finish_reason": "tool_calls"
    }
  ]
}

​
tool_choice Parameter

ValueDescription
"auto"Default, the model decides whether to call tools
"none"Disable tool calling
"required"Force the model to call at least one tool
{"type": "function", "function": {"name": "get_weather"}}Force calling a specific function

​
Parallel Function Calling

The model can request multiple function calls in a single response: 
{
  "tool_calls": [
    {
      "id": "call_001",
      "type": "function",
      "function": {
        "name": "get_weather",
        "arguments": "{\"city\": \"Beijing\"}"
      }
    },
    {
      "id": "call_002",
      "type": "function",
      "function": {
        "name": "get_weather",
        "arguments": "{\"city\": \"Shanghai\"}"
      }
    }
  ]
}
In this case, you need to return a corresponding tool message for each tool_call, and the tool_call_id must match.
function.arguments is a JSON string that needs to be parsed with JSON.parse() or json.loads(). The arguments generated by the model may not fully conform to the schema, so proper error handling is recommended.
Not all models support Function Calling. Make sure the model you are using supports this feature.