---
id: arviz
version: "1.3.0"
license: unclear
license_treatment: permissive
maintenance: active
---

# arviz — Expose features from _ArviZverse_ refactored packages together in the ``arviz`` namespace.

License: permissive · Maintenance: active · Downloads: 3.9M/mo

## What it is and what it does

ArviZ is a modular library for analyzing Bayesian models after inference. It sits downstream of sampling tools (like PyMC or Stan) and provides a unified interface for posterior diagnostics, visualization, and model comparison. The package exposes features from three refactored subpackages—arviz_base, arviz_stats, and arviz_plots—together in the arviz namespace, allowing users to work with inference results in a consistent way regardless of which sampler generated them.

The library is designed for exploratory analysis: inspecting posterior distributions, checking convergence, comparing competing models, and validating inference quality. It handles data storage and transformation, making it straightforward to load traces from different Bayesian frameworks and apply a common suite of diagnostics and plots. Development is active (last commit 2026-08-11) and the package supports modern Python versions.

Use it for:

- Inspect posterior distributions and convergence diagnostics after running MCMC or variational inference.
- Compare competing Bayesian models using built-in comparison metrics and visualizations.
- Validate inference quality and detect sampling issues through standardized diagnostic plots.
- Store and reload Bayesian inference results in a framework-agnostic format.
- Generate publication-quality plots of Bayesian analysis results for reports or papers.

## Worth the install?

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

ArviZ is a Python package for exploratory analysis of Bayesian models, providing functions for posterior analysis, data storage, model checking, comparison, and diagnostics.

Yes. ArviZ is a mature, actively maintained tool (release 3 days old, 1846 repository stars) with no known vulnerabilities, permissive licensing, and low install friction. It is essential for anyone working with Bayesian inference in Python who needs systematic posterior analysis and diagnostics beyond what individual samplers provide.

## Install

pip install arviz
uv add arviz
poetry add arviz

## Installing arviz

Before you install: Low friction installation as a pure Python wheel. Actively maintained with a release 3 days old and recent commits; supports current Python versions (3.12, 3.13, 3.14).

License in practice: Licensed under Apache (permissive), imposing no significant restrictions on use or redistribution.

Quickstart:

pip install arviz

import arviz as az
# Load and analyze Bayesian model results
data = az.from_pymc3(trace=your_trace, prior=your_prior)
az.plot_posterior(data)

Requires Python 3.12 or later.

Verify before relying:

- Whether arviz_base, arviz_stats, and arviz_plots are required at runtime or optional modular components.
- Specific Bayesian sampling frameworks (PyMC, Stan, etc.) that ArviZ integrates with beyond what the fact sheet states.

## Package facts

- License: not declared (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 3.9M/month (top 5,000 on PyPI)
- Known vulnerabilities: none known

## Tags

bayesian model analysis, posterior diagnostics, bayesian inference visualization, mcmc diagnostics, bayesian model comparison, bayesian-inference, diagnostics, visualization

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