How to Intermediate · 4 min read

How to use Mistral function calling

Q: How to use Mistral function calling

Use the mistralai Python SDK or the OpenAI-compatible openai SDK with base_url="https://api.mistral.ai/v1" to call Mistral models. Define functions in the functions parameter and handle the model's function call response to execute or simulate function calls.

Quick answer

Use the mistralai Python SDK or the OpenAI-compatible openai SDK with base_url="https://api.mistral.ai/v1" to call Mistral models. Define functions in the functions parameter and handle the model's function call response to execute or simulate function calls.

PREREQUISITES

Python 3.8+
MISTRAL_API_KEY environment variable set
pip install openai>=1.0 or mistralai

Setup

Install the openai package for OpenAI-compatible usage or the official mistralai SDK. Set your API key in the environment variable MISTRAL_API_KEY.

bash

pip install openai
# or
pip install mistralai

Step by step

Use the OpenAI-compatible openai SDK to call a Mistral model with function calling. Define the function schema in functions and check if the model requests a function call in the response. Then handle the function call accordingly.

python

import os
from openai import OpenAI

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

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 New York?"}
]

response = client.chat.completions.create(
    model="mistral-large-latest",
    messages=messages,
    functions=functions,
    function_call="auto"
)

choice = response.choices[0]
message = choice.message

if message.function_call:
    func_name = message.function_call.name
    func_args = message.function_call.arguments
    print(f"Function call requested: {func_name}")
    print(f"Arguments: {func_args}")
    # Here you would implement the function logic or call an API
    # For demo, simulate a response
    function_response = "It's sunny and 75 degrees Fahrenheit in New York."

    # Send the function response back to the model
    followup_response = client.chat.completions.create(
        model="mistral-large-latest",
        messages=[
            *messages,
            message,
            {"role": "function", "name": func_name, "content": function_response}
        ]
    )
    print(f"Model reply: {followup_response.choices[0].message.content}")
else:
    print(f"Model reply: {message.content}")

output

Function call requested: get_current_weather
Arguments: {"location": "New York"}
Model reply: It's sunny and 75 degrees Fahrenheit in New York.

Common variations

Use the official mistralai SDK with client.chat.complete() for function calling.
Set function_call="none" to disable function calling.
Use async calls with asyncio and await if supported.
Switch models like mistral-small-latest for faster responses.

python

import os
from mistralai import Mistral

client = Mistral(api_key=os.environ["MISTRAL_API_KEY"])

functions = [
    {
        "name": "get_current_weather",
        "description": "Get the current weather in a given location",
        "parameters": {
            "type": "object",
            "properties": {
                "location": {"type": "string"},
                "unit": {"type": "string", "enum": ["celsius", "fahrenheit"]}
            },
            "required": ["location"]
        }
    }
]

response = client.chat.complete(
    model="mistral-large-latest",
    messages=[{"role": "user", "content": "Weather in Boston?"}],
    functions=functions,
    function_call="auto"
)

print(response.choices[0].message.content)

output

Function call requested: get_current_weather
Arguments: {"location": "Boston"}
Model reply: It's currently 60 degrees Fahrenheit and cloudy in Boston.

Troubleshooting

If you get 401 Unauthorized, verify your MISTRAL_API_KEY is set correctly.
If function calls are not triggered, ensure functions parameter is correctly formatted as JSON schema.
Check network connectivity to https://api.mistral.ai/v1.
Use function_call="auto" to enable automatic function calling by the model.

✅

Key Takeaways

Use the OpenAI-compatible openai SDK with base_url="https://api.mistral.ai/v1" for Mistral function calling.
Define functions using JSON schema in the functions parameter and set function_call="auto" to enable calls.
Handle the model's function_call response to execute or simulate the requested function and send results back.
The official mistralai SDK offers similar function calling support with client.chat.complete().
Always secure your API key in environment variables and verify correct formatting to avoid common errors.

Verified 2026-04 · mistral-large-latest, mistral-small-latest

Verify ↗