Integrate Langtrace with Mistral

Karthik Kalyanaraman

⸱

Cofounder and CTO

Nov 14, 2024

Introduction

We are excited to announce that Langtrace now supports Mistral's family of models natively. This means Langtrace will automatically capture traces and metrics including token usage, cost, latency and model hyper parameters automatically if you are using Mistral's family of models.

Setup

Sign up to Langtrace, create a project and get a Langtrace API key
Install Langtrace SDK

pip install -U langtrace-python-sdk

Setup the .env var

export LANGTRACE_API_KEY=YOUR_LANGTRACE_API_KEY

Initialize Langtrace in your code

import os
from langtrace_python_sdk import langtrace, with_langtrace_root_span
from mistralai import Mistral

langtrace.init()

@with_langtrace_root_span("chat_complete")
def chat_complete():
    model = "mistral-large-latest"
    client = Mistral(api_key=os.environ["MISTRAL_API_KEY"])
    chat_response = client.chat.complete(
        model=model,
        messages=[
            {
                "role": "user",
                "content": "I need 10 cocktail recipes with tequila other than the classics like margarita, tequila"
            },
        ]
    )
    print(chat_response.choices[0].message.content)


chat_complete()