ZenML
Munich open source orchestration for ML pipelines and agents, with artifact tracking and a model control plane
Quick Overview
| Company | ZenML GmbH |
|---|---|
| Category | MLOps Platforms |
| Headquarters | Munich, Germany |
| Founded | Not stated |
| EU Presence | Yes - Germany (EU) |
| Data Location | Self-hosted on your infrastructure, or ZenML Pro SaaS |
| Open Source | Apache 2.0 |
| GDPR Compliant | Yes |
| Self-Hosting | Yes - open source, self-hosted |
| Main Features | Pipeline and flow orchestration, artifact management, basic model registry, Model Control Plane, snapshots, Codespaces, SSO and audit logs on higher plans |
| Pricing | Open source free; Scale SaaS $999/month; Enterprise on request |
| Best For | ML engineers who want pipelines to move between orchestrators and clouds |
| Replaces | Weights & Biases, Comet, Databricks MLflow |
Detailed Review
ZenML is aimed at the step before tracking dashboards: how a training script becomes a repeatable pipeline. You write steps in Python, and ZenML runs them on the orchestrator and cloud you already operate, so a pipeline written on a laptop can run on Kubernetes without a rewrite.
The company is ZenML GmbH, registered in Munich (HRB 268487), and its terms are governed by German law. The home page lists SOC 2 Type II and ISO 27001, and the core is published under Apache 2.0.
Pipelines first, tracking second
The open source plan lists pipeline and flow orchestration, artifact management and a basic model registry. Every pipeline run stores its artifacts and metadata, which gives you lineage between a model and the data that produced it.
Teams who currently use Weights & Biases for run charts should read the feature table carefully, because the pricing page names orchestration and artifacts rather than a chart-comparison dashboard.
What the paid plans add
Scale SaaS costs $999 per month and bundles 2,000 executions, 3 projects and 5 snapshots, with the Model Control Plane, Artifact Control Plane and Codespaces. Execution tiers of 500, 2,000 and 5,000 are shown. Enterprise adds SSO, custom roles, audit logs and air-gapped deployment.
The pricing page shows prices in dollars, so budget for the conversion if you pay in euros.
Agents and Kitaru
ZenML now sits beside Kitaru, a separate product for recording and replaying agent sessions on production data. Kitaru Cloud is a flat $39 per month for 3 agents and 2 seats, with a 14-day trial, and its core is also Apache 2.0.
If you only need classic ML pipelines, Kitaru can be ignored, but the two share a subscription and a workspace.
Limitations to Consider
ZenML orchestrates and records; it does not ship the dense run-comparison dashboards that Weights & Biases is known for, so pair it with DagsHub or MLflow if you want those. The open source model registry is described as basic, and model governance features sit in the paid Pro control plane. Scale SaaS at $999 per month is a large step from the free plan. You still run the orchestrator and compute yourself.
Who Should Use ZenML
ML engineering teams in Europe that want an Apache 2.0 pipeline layer from a German vendor, run on their own Kubernetes or cloud accounts.
Pros and Cons
Pros
- Apache 2.0 core that you can self-host
- Pipelines move between orchestrators without rewrites
- German company with German governing law
- SOC 2 Type II and ISO 27001 listed
- Free open source plan with no execution cap
Cons
- No chart-heavy run comparison dashboard on the plans listed
- Scale SaaS jumps to $999 per month
- Prices shown in dollars
- Basic model registry in the open source edition
Alternatives to ZenML
Other European mlops platforms worth comparing before you decide:
Valohai
Finland · Turku platform for versioned ML runs, pipelines and GPU scheduling
Hopsworks
Sweden · Stockholm feature store with model registry and serving, free tier
DagsHub
Israel · Experiment tracking, data versioning and annotation on one platform
UbiOps
Netherlands · Hague platform to deploy and orchestrate models on any infrastructure
See all European mlops platforms or compare them directly against Weights & Biases on our alternatives page.
Frequently Asked Questions
Partly. ZenML covers pipelines, artifacts and a model registry, while the dashboards for comparing many runs come from tools such as DagsHub or MLflow that it integrates with.
Yes. ZenML and Kitaru are Apache 2.0, and the open source plan is described as self-hosted on your own infrastructure with no cap on executions.
The Scale SaaS plan is $999 per month on 23 October 2026, Enterprise is custom, and the Kitaru Cloud agent product is a flat $39 per month.