What PyMC does

PyMC is an open-source probabilistic programming library written in Python that provides a high-level interface for specifying Bayesian statistical models and running inference via Markov Chain Monte Carlo (NUTS, HMC) and variational methods. The library is licensed under Apache 2.0 and distributed through GitHub, PyPI, conda-forge, and Google Colab; the current release is 6.2.0. Its computational backend was migrated from Theano to PyTensor, with optional JAX compilation that enables execution on CPUs, GPUs, and TPUs. PyMC integrates tightly with adjacent scientific-Python tooling — ArviZ for diagnostics and visualization, Bambi for generalized linear mixed models, and JAX for accelerated numerics — forming a coherent stack for applied Bayesian modeling.

The primary user base consists of academic researchers, statisticians, and applied data scientists working in domains where uncertainty quantification matters, including pharmacology, epidemiology, ecology, marketing science, and the social sciences. The project is community-led and fiscally sponsored by NumFOCUS, a 501(c)(3) nonprofit that also channels donations.

PyMC is not a venture-backed company and has no disclosed headcount, funding, or customer roster. Monetization is indirect: the core library is free, and the only identifiable revenue surface is PyMC Labs, a separate consulting entity staffed by core contributors that sells Bayesian modeling services — including a packaged Bayesian Media Mix Models offering — and accepts donations via NumFOCUS. The go-to-market is therefore community-led on the open-source side and boutique-consulting-led on the commercial side, with no enterprise sales motion disclosed.

Name

PyMC

Website

https://docs.pymc.io

Company type

Private

Founded year

2015

Operating status

Operating

Headcount range

1–10 employees

Short description

PyMC is an open-source Python probabilistic programming library for Bayesian statistical modeling, serving academic researchers and applied data scientists; the core project is free, with monetization occurring indirectly through PyMC Labs consulting and NumFOCUS donations.

Ownership category

akta.pro rank

GTM type

B2B and B2C

Offering type

Software

Cost components

Personnel, Technology or R&D, Operations, Marketing or Sales, Others

Revenue model

1. Open Source Distribution: PyMC is freely distributed as open-source software under the Apache License 2.0. Revenue is not generated directly from the software. The project is supported by donations through NumFOCUS.
2. Consulting Services (PyMC Labs): PyMC Labs, a separate Bayesian consultancy consisting of PyMC core development team members, offers professional consulting services including model speed-ups, improving models, building custom models for business problems, and Bayesian Media Mix Models for marketing attribution. Contact email: [email protected].

Go-to-market motion1 record

Distribution channels5 records

Marketing channels8 records

Core offering

PyMC is an open-source Python library for Bayesian statistical modeling and probabilistic programming, providing MCMC sampling algorithms (NUTS, HMC, Metropolis-Hastings), probability distributions, Gaussian process models, and generalized linear models. Built on PyTensor for symbolic computation and automatic differentiation, it enables an end-to-end Bayesian workflow from model specification through inference, diagnostics, and model comparison. A separate consultancy, PyMC Labs, offers paid professional services staffed by core team members for custom model development and optimization.

Product overview

PyMC is a single unified open-source Python library for Bayesian statistical modeling and Probabilistic Programming. The core product is the PyMC library itself, which provides MCMC sampling (NUTS, HMC, Metropolis-Hastings), probability distributions, Gaussian Process modeling (with multiple covariance functions including ExpQuad, Matern, Periodic, and additive GP support), Generalized Linear Models, and integration with PyTensor for computation. The ecosystem includes PyMC Labs as a professional consulting service staffed by core team members for model speed-ups, custom model development, and Bayesian Media Mix Models. An experimental pmx library provides cutting-edge functionality beyond the core library. PyMC integrates with Bambi (for formula-based GLM specification) and ArviZ (for MCMC diagnostics and visualization), and outputs results in xarray DataTree format.

Differentiator

Problem solved

Functional benefit

Products and services

• PyMC (Core Library) Open-source Python library for Bayesian statistical modeling and Probabilistic Programming, providing MCMC sampling algorithms (NUTS, HMC, Metropolis-Hastings, DEMetropolis), probability distributions, Gaussian Process modeling with an extensive covariance function library, Generalized Linear Models, and full Bayesian workflow tooling. Built on PyTensor for automatic differentiation and symbolic computation, it targets data scientists, statisticians, and researchers needing Python-native Bayesian inference.
• PyMC Labs Consulting Services Paid Bayesian consulting services staffed by PyMC core development team members, including model speed-ups (reparameterizations, JAX, GPU sampling), model improvements (adding hierarchy, time-series structure), custom model development for applied business problems, and Bayesian Media Mix Models for marketing attribution. Targets enterprise organizations needing expert implementation help.

Segments3 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration3 records

AI capability3 records

Feature9 records

Partnerships

Two partnerships are on record, tiered core and minor.

• PyMC LabscoreImplementation/ SI/ Consulting PartnerPyMC Labs is a Bayesian consultancy consisting of members of the PyMC core development team. They offer professional services including model speed-ups (reparameterizations, JAX, GPU sampling), improving models (adding hierarchy, time-series structure), building new models for applied business problems, and Bayesian Media Mix Models for marketing attribution. Contact: [email protected].
• NumFOCUSminorOthersNumFOCUS is a nonprofit organization that fiscal hosts the PyMC project and handles donations. Users can support PyMC through the NumFOCUS donation page at numfocus.org/donate-to-pymc.

Scale indicators3 records

Recent moves6 records

Expansion highlights5 records

Direct peers

• Stan: Stan is the leading probabilistic programming language for Bayesian inference, offering Hamiltonian Monte Carlo sampling with its own modeling language. It is the most direct functional competitor to PyMC, sharing the same target users (statisticians, data scientists) and core value proposition.
• Pyro: Pyro is a universal probabilistic programming language built on PyTorch, originally developed at Uber. It serves the same Bayesian modeling audience as PyMC with deep-learning-friendly primitives, making it a direct competitor for users coming from the PyTorch ecosystem.
• JAGS: JAGS (Just Another Gibbs Sampler) is a legacy Bayesian modeling program using Gibbs sampling. It targets the same statistical user base as PyMC for hierarchical Bayesian models but with an older, less Pythonic interface that PyMC explicitly set out to displace.
• BUGS / OpenBUGS: OpenBUGS is an open-source Bayesian inference framework using MCMC methods, representing the traditional academic Bayesian software lineage. It shares PyMC's core value proposition of making Bayesian methods accessible but uses a different (declarative) modeling language.
• NumPyro: NumPyro is a JAX-based probabilistic programming library offering similar Bayesian inference capabilities to PyMC with native GPU/TPU acceleration. It competes head-to-head on performance-sensitive use cases and represents the leading performance-focused alternative.

Others

• Bambi: Bambi is a Python library for building generalized linear models using formula syntax on top of PyMC. It is a sibling project in the PyMC ecosystem that lowers the barrier to PyMC for non-statisticians, complementing rather than competing with the core library.
• ArviZ: ArviZ is the diagnostics and visualization library tightly integrated with PyMC for MCMC analysis. While not a direct competitor, it is a core part of the PyMC ecosystem and demonstrates how the project has built an interoperable stack rather than competing alone.
• statsmodels: statsmodels is a Python module providing classes and functions for statistical models including Bayesian methods. It is adjacent to PyMC with overlapping user base but a broader frequentist focus, sometimes used alongside PyMC for frequentist initialization (via Bambi).

Emerging players

• Edward / Edward2: Edward was a probabilistic programming library built on TensorFlow (now succeeded by Edward2 under DeepMind). It targets similar use cases as PyMC with deep-learning integration, focusing on variational inference alongside MCMC methods.

Broad incumbents

• TensorFlow Probability: TensorFlow Probability is Google's library for probabilistic reasoning and statistical analysis built on TensorFlow. As a broader incumbent backed by a major platform, it competes with PyMC for users in the TensorFlow ecosystem while offering wider scope beyond pure Bayesian inference.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks6 records

Key highlights7 records

Customer concentration

• Xx.com/pymc_devs
• GitHubgithub.com/pymc-devs/pymc
• YouTubeyoutube.com/c/PyMCDevelopers
• LinkedInlinkedin.com/company/pymc

Revenue estimate

Valuation estimate

Number of profiles

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about PyMC