How to intermediate · 3 min read

Fix LLM not calling function when expected

Q: Fix LLM not calling function when expected

Ensure you use the tools parameter (not deprecated functions) when creating chat completions with the OpenAI SDK. Also, check if the response's finish_reason is tool_calls and parse message.tool_calls to trigger the function call properly.

Quick answer

Ensure you use the tools parameter (not deprecated functions) when creating chat completions with the OpenAI SDK. Also, check if the response's finish_reason is tool_calls and parse message.tool_calls to trigger the function call properly.

PREREQUISITES

Python 3.8+
OpenAI API key (free tier works)
pip install openai>=1.0

Setup

Install the latest openai Python package and set your API key as an environment variable.

Install SDK: pip install openai --upgrade
Set environment variable: export OPENAI_API_KEY='your_api_key' (Linux/macOS) or setx OPENAI_API_KEY "your_api_key" (Windows)

bash

pip install openai --upgrade

output

Collecting openai
  Downloading openai-1.x.x-py3-none-any.whl (xx kB)
Installing collected packages: openai
Successfully installed openai-1.x.x

Step by step

Use the tools parameter to define your function schema and check the response for tool_calls to invoke the function. This example shows a weather function call triggered by the LLM.

python

import os
import json
from openai import OpenAI

client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])

# Define the function schema as a tool
weather_tool = [{
    "type": "function",
    "function": {
        "name": "get_weather",
        "description": "Get weather for a location",
        "parameters": {
            "type": "object",
            "properties": {
                "location": {"type": "string"}
            },
            "required": ["location"]
        }
    }
}]

# User prompt
messages = [{"role": "user", "content": "What's the weather in New York?"}]

# Create chat completion with tools
response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=messages,
    tools=weather_tool
)

print("LLM response:", response.choices[0].message.content)

# Check if LLM wants to call a function
if response.choices[0].finish_reason == "tool_calls":
    tool_call = response.choices[0].message.tool_calls[0]
    args = json.loads(tool_call.function.arguments)
    location = args.get("location")
    # Simulate function call
    weather_result = f"The weather in {location} is sunny and 75°F."
    print("Function call result:", weather_result)
else:
    print("No function call requested.")

output

LLM response: 
Function call result: The weather in New York is sunny and 75°F.

Common variations

You can use async calls with the OpenAI SDK by using asyncio and await. Also, you can switch to different models like gpt-4o or gpt-4o-mini. Remember to always use the tools parameter instead of the deprecated functions parameter.

python

import os
import json
import asyncio
from openai import OpenAI

async def main():
    client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])

    weather_tool = [{
        "type": "function",
        "function": {
            "name": "get_weather",
            "description": "Get weather for a location",
            "parameters": {
                "type": "object",
                "properties": {"location": {"type": "string"}},
                "required": ["location"]
            }
        }
    }]

    messages = [{"role": "user", "content": "What's the weather in San Francisco?"}]

    response = await client.chat.completions.create(
        model="gpt-4o",
        messages=messages,
        tools=weather_tool
    )

    print("LLM response:", response.choices[0].message.content)

    if response.choices[0].finish_reason == "tool_calls":
        tool_call = response.choices[0].message.tool_calls[0]
        args = json.loads(tool_call.function.arguments)
        location = args.get("location")
        weather_result = f"The weather in {location} is cloudy and 65°F."
        print("Function call result:", weather_result)
    else:
        print("No function call requested.")

asyncio.run(main())

output

LLM response: 
Function call result: The weather in San Francisco is cloudy and 65°F.

Troubleshooting

If the LLM never calls your function, verify you are using the tools parameter instead of the deprecated functions parameter.
Check that your function schema is correctly formatted as a tool with type and function keys.
Ensure you parse response.choices[0].message.tool_calls and not response.choices[0].message.function_call, which is deprecated.
If finish_reason is not tool_calls, the model did not request a function call; try rephrasing your prompt or adjusting model temperature.

✅

Key Takeaways

Always use the tools parameter to define functions for the LLM to call.
Check finish_reason == 'tool_calls' and parse message.tool_calls to handle function calls.
Avoid deprecated parameters like functions and function_call in the OpenAI SDK.
Use async calls for non-blocking function invocation with the OpenAI Python SDK.
Validate your function schema strictly matches the required JSON schema format.

Verified 2026-04 · gpt-4o-mini, gpt-4o

Verify ↗