Function & Tool Calling Guide

Overview

InfinityRouter passes your tools definitions to capable models (such as Claude 3.5 Sonnet, GPT-4o, and DeepSeek-V3). When the model decides to invoke a tool, it returns a tool_calls array containing the function name and parsed JSON arguments.

1. Defining Tools

Tools are specified as an array of JSON Schema objects inside the tools parameter:

jsonTool Schema Definition

[
  {
    "type": "function",
    "function": {
      "name": "get_stock_quote",
      "description": "Fetch real-time ticker price and market cap for a stock symbol",
      "parameters": {
        "type": "object",
        "properties": {
          "symbol": {
            "type": "string",
            "description": "Stock ticker symbol, e.g. NVDA, AAPL"
          },
          "currency": {
            "type": "string",
            "enum": ["USD", "EUR"],
            "default": "USD"
          }
        },
        "required": ["symbol"]
      }
    }
  }
]

2. Complete Multi-Turn Tool Execution Loop (Python)

Here is a complete, self-contained Python script showing the request, the tool call interception, executing the local function, and returning the result for final synthesis:

pythontool_agent.py

import json
import os
from openai import OpenAI

client = OpenAI(
    base_url=os.environ.get("INFINITY_BASE_URL", "https://infinityrouter.qd.je/v1"),
    api_key=os.environ.get("INFINITY_API_KEY"),
)

# 1. Define local Python tool
def get_stock_quote(symbol: str, currency: str = "USD") -> str:
    mock_prices = {"NVDA": 132.50, "AAPL": 224.10, "GOOGL": 178.90}
    price = mock_prices.get(symbol.upper(), 100.00)
    return json.dumps({"symbol": symbol.upper(), "price": price, "currency": currency})

tools = [
    {
        "type": "function",
        "function": {
            "name": "get_stock_quote",
            "description": "Fetch real-time ticker price for a stock symbol",
            "parameters": {
                "type": "object",
                "properties": {
                    "symbol": {"type": "string", "description": "Stock ticker symbol"},
                    "currency": {"type": "string", "enum": ["USD", "EUR"], "default": "USD"}
                },
                "required": ["symbol"]
            }
        }
    }
]

messages = [
    {"role": "user", "content": "What is the current stock price of NVDA and AAPL?"}
]

# 2. First call: Model determines tool calls
response = client.chat.completions.create(
    model="claude-sonnet-5",
    messages=messages,
    tools=tools,
    tool_choice="auto",
)

assistant_message = response.choices[0].message
messages.append(assistant_message)

# 3. Process each tool call requested by the model
if assistant_message.tool_calls:
    for tool_call in assistant_message.tool_calls:
        fn_name = tool_call.function.name
        fn_args = json.loads(tool_call.function.arguments)
        
        if fn_name == "get_stock_quote":
            result = get_stock_quote(**fn_args)
            messages.append({
                "role": "tool",
                "tool_call_id": tool_call.id,
                "content": result,
            })

    # 4. Final call: Model synthesizes tool output into natural response
    final_response = client.chat.completions.create(
        model="claude-sonnet-5",
        messages=messages,
    )
    print(final_response.choices[0].message.content)