How to beginner · 3 min read

Fireworks AI function calling

Quick answer

Use the OpenAI-compatible Python SDK with Fireworks AI by specifying the tools parameter for function calling in chat.completions.create. Detect finish_reason == 'tool_calls' in the response to handle tool invocation and parse arguments from message.tool_calls.

PREREQUISITES

Python 3.8+
Fireworks AI API key
pip install openai>=1.0

Setup

Install the openai Python package (v1 or later) and set your Fireworks AI API key as an environment variable.

Install SDK: pip install openai
Set environment variable: export FIREWORKS_API_KEY='your_api_key' (Linux/macOS) or set FIREWORKS_API_KEY=your_api_key (Windows)

bash

pip install openai

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

This example shows how to call a function using Fireworks AI's OpenAI-compatible API. Define the tools parameter describing the function, send a chat completion request, and handle the tool call response.

python

import os
import json
from openai import OpenAI

client = OpenAI(api_key=os.environ["FIREWORKS_API_KEY"],
                base_url="https://api.fireworks.ai/inference/v1")

# Define the function as a tool
tools = [{
    "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 New York?"}]

response = client.chat.completions.create(
    model="accounts/fireworks/models/llama-v3p3-70b-instruct",
    tools=tools,
    messages=messages
)

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

# Check if the model requested a tool call
if response.choices[0].finish_reason == "tool_calls":
    tool_call = response.choices[0].message.tool_calls[0]
    function_name = tool_call.function.name
    arguments = json.loads(tool_call.function.arguments)
    print(f"Function called: {function_name}")
    print(f"Arguments: {arguments}")

output

Assistant reply: 
Function called: get_weather
Arguments: {'location': 'New York'}

Common variations

You can use different Fireworks AI models by changing the model parameter. Async calls are supported with async and await. Streaming responses are not currently supported for function calling. Always use the tools parameter instead of deprecated functions.

python

import asyncio

async def async_function_call():
    client = OpenAI(api_key=os.environ["FIREWORKS_API_KEY"],
                    base_url="https://api.fireworks.ai/inference/v1")

    tools = [{
        "type": "function",
        "function": {
            "name": "get_time",
            "description": "Get current time",
            "parameters": {"type": "object", "properties": {}, "required": []}
        }
    }]

    messages = [{"role": "user", "content": "What time is it?"}]

    response = await client.chat.completions.acreate(
        model="accounts/fireworks/models/llama-v3p3-70b-instruct",
        tools=tools,
        messages=messages
    )

    print("Async assistant reply:", response.choices[0].message.content)

asyncio.run(async_function_call())

output

Async assistant reply:

Troubleshooting

If you get an authentication error, verify your FIREWORKS_API_KEY environment variable is set correctly.
If finish_reason is not tool_calls, the model did not invoke the function; check your tools definition and prompt.
Use the correct Fireworks AI model name starting with accounts/fireworks/models/.

✅

Key Takeaways

Use the OpenAI-compatible tools parameter for function calling with Fireworks AI.
Check finish_reason == 'tool_calls' to detect when the model requests a function call.
Always parse message.tool_calls to extract function name and arguments.
Use environment variables for API keys and specify Fireworks AI base_url in the OpenAI client.
Avoid deprecated parameters like functions= or function_call=.

Verified 2026-04 · accounts/fireworks/models/llama-v3p3-70b-instruct

Verify ↗