Mistral → OpenObserve

Automatically capture token usage, latency, and model metadata for every Mistral AI inference call in your Python application.

Prerequisites

Python 3.9+
A Mistral account with an API key
An OpenObserve account (cloud or self-hosted)
Your OpenObserve organisation ID and Base64-encoded auth token

Installation

pip install openobserve-telemetry-sdk "openinference-instrumentation-mistralai==1.4.0" mistralai python-dotenv

Configuration

Create a .env file in your project root:

OPENOBSERVE_URL=https://api.openobserve.ai/
OPENOBSERVE_ORG=your_org_id
OPENOBSERVE_AUTH_TOKEN=Basic <your_base64_token>
MISTRAL_API_KEY=your-mistral-api-key

Instrumentation

Call MistralAIInstrumentor().instrument() before importing the Mistral client.

from dotenv import load_dotenv
load_dotenv()

from openinference.instrumentation.mistralai import MistralAIInstrumentor
from openobserve import openobserve_init

MistralAIInstrumentor().instrument()
openobserve_init()

import os
from mistralai import Mistral

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

response = client.chat.complete(
    model="mistral-small-latest",
    messages=[{"role": "user", "content": "Explain observability in one sentence."}],
)
print(response.choices[0].message.content)

What Gets Captured

Attribute	Description
`llm_model_name`	Model used (e.g. `mistral-small-latest`)
`gen_ai_response_model`	Model that served the response
`llm_system`	`mistralai`
`llm_provider`	`mistralai`
`llm_usage_tokens_input`	Prompt token count
`llm_usage_tokens_output`	Completion token count
`llm_usage_tokens_total`	Total tokens consumed
`openinference_span_kind`	`LLM`
`operation_name`	`MistralClient.chat`
`span_status`	`OK` or `ERROR`
`status_message`	Error details if the request failed
`duration`	Request latency

Viewing Traces

Log in to OpenObserve and navigate to Traces in the left sidebar
Filter by llm_model_name to find Mistral spans
Click any span to inspect token counts, model name, and latency
Error spans show span_status = ERROR with the full error message in status_message

Mistral trace in OpenObserve

Next Steps

With Mistral instrumented, every inference call is recorded in OpenObserve. From here you can compare latency across Mistral models, track token consumption over time, and set alerts on error rates.