---
id: arthur-client
version: "1.4.2333"
license: MIT
license_treatment: permissive
maintenance: active
---

# arthur-client — Arthur Python API Client Library

License: permissive · Maintenance: active · Downloads: 96.5K/mo

## What it is and what it does

Arthur-client is a Python SDK that connects your machine learning models to Arthur's centralized monitoring platform. It abstracts the HTTP API calls needed to send model metadata, predictions, and performance metrics to Arthur's infrastructure, enabling teams to track model accuracy, explainability, and fairness across production deployments without being locked into a specific ML framework or infrastructure provider.

The package depends on standard HTTP and configuration libraries (requests, urllib3, authlib, pydantic, click) to handle authentication, data validation, and CLI interactions. It's designed for data scientists and ML engineers who want to integrate monitoring into their model pipelines with minimal setup overhead—typically just instantiating a client and calling methods to register models and log predictions.

Use it for:

- Integrate a scikit-learn or neural network model into Arthur's monitoring dashboard to track real-time prediction accuracy and data drift.
- Send model fairness metrics to Arthur for compliance auditing and bias detection across demographic groups.
- Set up automated alerts when model performance degrades below defined thresholds in production.
- Log prediction explanations and feature importance alongside predictions for model interpretability.
- Centralize monitoring for multiple models deployed across different services or cloud providers.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Python SDK for integrating machine learning models with Arthur, a centralized platform for monitoring production models across accuracy, explainability, and fairness.

Yes, if you are already using or planning to adopt Arthur for model monitoring. The SDK is actively maintained, has low install friction, uses a permissive MIT license, and carries no known vulnerabilities. Install only if your team has committed to Arthur's platform; it is a client library for a specific service, not a general-purpose monitoring tool.

## Install

pip install arthur-client
uv add arthur-client
poetry add arthur-client

## Installing arthur-client

Before you install: Low install friction with a pure-Python wheel distribution. Actively maintained as of the latest release date. Supports current Python versions (3.12, 3.13) with no compiled dependencies.

License in practice: MIT license permits commercial and private use with minimal restrictions, making it suitable for most production deployments without legal friction.

Quickstart:

pip install arthur-client

from arthur_client import ArthurClient

client = ArthurClient()
# Configure and connect your model to Arthur platform

Requires Python 3.12 or later.

Verify before relying:

- Specific authentication mechanisms and credential handling beyond authlib dependency
- Whether the SDK supports all Arthur platform features or a subset of the API
- Performance characteristics when monitoring high-volume model predictions
- Integration patterns with common ML frameworks (scikit-learn, TensorFlow, PyTorch)

## Package facts

- License: MIT (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 96.5K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

ml model monitoring sdk, production model observability, arthur platform integration, model performance tracking, ai model governance, ml ops monitoring client, model fairness tracking, ml-monitoring, model-observability, arthur-platform

[View on SkillFed](https://skillfed.io/packages/arthur-client) · [View on PyPI](https://pypi.org/project/arthur-client/)