Flexible Deployment for Your MLOps Needs

Streamline your machine learning workflows with ZenML's adaptable deployment options. Whether you're experimenting locally or scaling to production, ZenML empowers you to orchestrate, manage, and deploy your models with ease.
Try Free
Read the Documentation
ZenML Architecture

Built on a Robust Client-Server Architecture

ZenML is a metadata layer on top of your existing infrastructure, meaning all data and compute stays on your side.
ZenML system architecture diagram showing connections between five main components: ZenML Client (Development Environment), ZenML Server, Database, MLOps Infrastructure (Cloud, Kubernetes, on-prem), and MLOps Tools (Experiment tracker, model deployer)
ZenML Server

Central hub managing your ML metadata

Store your pipelines, runs, stacks, and configurations. It provides a REST server for communication.
ZenML Client

Your interface to interact with the ZenML Server

It enables you to trigger actions, access information, and control your ML operations.
ZenML Dashboard

Visualize your pipelines, artifacts, and stacks.

Usually bundled with the server, the dashboard provides a seamlesss visual experience on top of the ZenML Client.
Deployments comparison

Choose the Deployment that Fits Your Needs

Feature
Open Source (Self-hosted)
SaaS
On-Prem
Run pipelines on cloud infra
Managed platform
No setup required
Easy version upgrade
RBAC
Model Control Plane
Artifact Control Plane
Run Templates
SSO / Auth2
No metadata leaves customer cloud

Get started quickly with Local Deployment

  • Ideal for initial experimentation, learning ZenML, and developing pipelines.
  • Easy setup - everything runs on your machine.
  • Uses a local SQLite database for storing metadata.
Learn More
A diagram showing the ZenML Default Local Configuration architecture. The image displays three main components: on the left, two boxes labeled 'User Code' and 'CLI' connect via bidirectional arrows to a central box containing the 'ZenML Client'. On the right, the ZenML Client connects to a 'Metadata Store' labeled with '(SQLite)'. The diagram has a purple and teal color scheme, with the ZenML logo and 'OSS' badge in the top-right corner.
Dashboard mockup

Deploy the ZenML server to start collaborating

  • Deploy the ZenML server and dashboard centrally as soon as you are deploying remote workloads or inviting colleagues
  • Backed by a robust database (MySQL) to persist production workloads
  • Suitable for production-level workloads.
Learn More

Easily upgrade to ZenML Pro to unlock enterprise -grade features

  • Best for teams and organizations looking to scale MLOps
  • Pro adds on top of OSS with advanced functionality and enterprise controls
  • Seamlessly upgrade to ZenML Pro from OSS with one-click
Learn More
A diagram showing the ZenML Pro Deployment architecture. The layout consists of a ZenML Client on the left connecting to multiple components. The deployment includes a Pro Control Plane at the top containing a ZenML Pro Server, Pro Metadata Store (PostgreSQL), and Pro Dashboard (Javascript). Below are two identical ZenML Pro Tenants (1 and 2), each labeled as 'OSS Tenant + Pro Features' and containing four components: ZenML OSS Server (Python), Pro Add-ons (Python), OSS Metadata Store (MySQL), and a Secrets Store. An Identity Provider component connects to the system through the Control Plane. The diagram uses purple for core components, orange for Pro features, and teal for storage elements. The ZenML logo with 'PRO' badge appears in the deployment and control plane headers.
Various methods available:
Purple torii gate icon on a circular background, symbolizing ZenML's streamlined machine learning pipelines.

ZenML CLI

Simplifies deployment on managed Kubernetes services.
Docker

Docker

Run ZenML Server in a container, optionally connecting to an external MySQL database.
Kubernetes

Helm

Deploy on a Kubernetes cluster using Helm charts for robust, production-ready setups.
Friendly yellow emoji with open arms and a smiley face.

Hugging Face Spaces

Quick and free deployment option, ideal for trying out ZenML.
ZenML architecture

Focus on What Matters - Your MLOps

ZenML is a metadata layer on top of your existing infrastructure, meaning all data and compute stays on your side.
Local
OSS
A FastAPI-based server managing metadata of pipelines, artifacts, and stacks. Free with Apache 2.0 license, perfect for individual projects and small teams starting their MLOps journey.
SaaS
PRO
Fully managed ZenML deployment with enhanced collaboration features, RBAC, and enterprise support. Ideal for teams looking for a zero-maintenance MLOps solution.
Self-hosted
PRO
Enterprise-grade deployment on your infrastructure. Complete control over data and secrets with all Pro features. Perfect for organizations with strict security requirements.
ZenML OSS deployment architecture showing two user environments (User 1 and User 2), each containing User Code and CLI components connected to a ZenML Client. Both clients connect to a central ZenML Deployment containing a ZenML OSS Server (Python), OSS Dashboard (Javascript), and Metadata Store (MySQL).
ZenML Pro Deployment architecture diagram showing a control plane with Identity Provider, ZenML Pro Server, Pro Metadata Store (PostgreSQL), and Pro Dashboard (Javascript). The diagram illustrates two tenants (ZenML Pro Tenant 1 and 2) labeled as OSS Tenant + Pro Features, each containing ZenML OSS Server (Python), Pro Add-ons (Python), OSS Metadata Store (MySQL), and Secrets Store. A ZenML Client connects to both the Identity Provider and tenant infrastructure.
Customer Environment architecture diagram for ZenML Pro showing core components: Pro Metadata Store (PostgreSQL), Pro Dashboard, and ZenML Control Plane. The environment includes two ZenML Pro Tenants connected to external services like Container Registry, Secrets Store, Data Lake, Artifact Store, Orchestrator, and Experiment Tracker. An Identity Provider sits above the environment, and a ZenML Client connects to Tenant 1 from below.

Fully managed service

ZenML handles deployment, maintenance, and infrastructure.

Simplified workflow

Allows you to focus on your ML workflows without infrastructure management overhead.

Scalable architecture

Suitable for teams and organisations looking for a streamlined and scalable solution.

Full on-prem available

Full on-prem deployment available as well (ask for details)

Enhanced observability

Enhanced security features and integrations with cloud services.

Risk-free evaluation

14-day free trial for thorough testing and POC development.

Start Your Free Trial Now

No new paradigms - Bring your own tools and infrastructure
No data leaves your servers, we only track metadata
Free trial included - no strings attached, cancel anytime
Try Free
Book a Demo
Dashboard displaying machine learning models, including versions, authors, and tags. Relevant to model monitoring and ML pipelines.
By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.
PreferencesDenyAccept