How to intermediate · 4 min read

How to use function calling with OpenAI Assistants API

Q: How to use function calling with OpenAI Assistants API

Use the OpenAI Assistants API by specifying the functions parameter in your chat.completions.create call to define callable functions. The assistant can then respond with a function_call object, which your code can parse and execute accordingly.

Quick answer

Use the OpenAI Assistants API by specifying the functions parameter in your chat.completions.create call to define callable functions. The assistant can then respond with a function_call object, which your code can parse and execute accordingly.

PREREQUISITES

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

Setup

Install the official OpenAI Python SDK and set your API key as an environment variable.

Install SDK: pip install openai
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

Step by step

This example demonstrates defining a function schema, sending it with the chat request, and handling the assistant's function call response.

python

import os
from openai import OpenAI

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

# Define the function schema
functions = [
    {
        "name": "get_current_weather",
        "description": "Get the current weather in a given location",
        "parameters": {
            "type": "object",
            "properties": {
                "location": {
                    "type": "string",
                    "description": "The city and state, e.g. San Francisco, CA"
                },
                "unit": {
                    "type": "string",
                    "enum": ["celsius", "fahrenheit"]
                }
            },
            "required": ["location"]
        }
    }
]

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

response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=messages,
    functions=functions,
    function_call="auto"  # Let the model decide to call the function
)

message = response.choices[0].message

if message.get("function_call"):
    function_name = message["function_call"]["name"]
    arguments = message["function_call"]["arguments"]
    print(f"Function to call: {function_name}")
    print(f"Arguments: {arguments}")
    # Here you would parse arguments and call your local function
else:
    print(message.content)

output

Function to call: get_current_weather
Arguments: {"location": "Boston"}

Common variations

Explicit function call: Set function_call="get_current_weather" to force the model to call a specific function.
Async usage: Use async client methods if your environment supports async.
Different models: Use gpt-4o-mini or gpt-4o depending on latency and cost needs.

python

import asyncio
import os
from openai import OpenAI

async def main():
    client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])
    response = await client.chat.completions.acreate(
        model="gpt-4o-mini",
        messages=[{"role": "user", "content": "Tell me the weather in New York."}],
        functions=functions,
        function_call="auto"
    )
    message = response.choices[0].message
    print(message)

asyncio.run(main())

Troubleshooting

If you receive an error about missing functions parameter, ensure you pass the function schema correctly as a list of dicts.
If the assistant never calls your function, try setting function_call="auto" explicitly or check your function descriptions for clarity.
Check your API key environment variable if authentication errors occur.

✅

Key Takeaways

Use the functions parameter to define callable functions in your chat request.
Check the function_call field in the assistant's response to trigger local function execution.
Set function_call="auto" to let the model decide when to call functions.
Use async client methods for non-blocking calls in async environments.
Ensure your function schemas are clear and correctly formatted to improve function calling accuracy.

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

Verify ↗