ZenML

What is ZenML?

ZenML is an open-source platform for orchestrating AI workflows written in Python. Pipelines cover training, inference, evaluation or agent tasks, run on infrastructure such as Kubernetes, Vertex, SageMaker or AzureML, and ZenML acts as a metadata layer on top, so data and compute stay in your own cloud. Alongside it, Kitaru records every agent run and lets you replay it against the recorded session, so you can test a code change against past sessions before release. The open-source edition is self-hosted and free. Hosted options include Kitaru Cloud at $39/month with 3 agents, 2 seats and 90-day session retention, and ZenML Scale at $999/month for teams running ML in production, with Enterprise tiers for SSO, audit logs and air-gapped deployment. ZenML Pro covers both workspaces under one organization, with roles, SSO and a managed control plane.

Use cases

• Orchestrate ML training and inference pipelines
• Replay agent runs against recorded sessions
• Track artifacts and models across runs
• Run pipelines on Kubernetes, Vertex or SageMaker
• Import traces from Langfuse or LangSmith for replay
• Set up evals for Python and TypeScript agents
• Give a team a managed control plane for AI workflows

Who it's for

• ML engineers
• MLOps teams
• AI engineers
• Data science teams

Why use ZenML

• The same pipelines run in the open-source and paid versions, so moving to Pro needs no code rewrite.
• ZenML is a metadata layer, which means your data and compute remain in your own infrastructure.
• Kitaru re-executes your agent code against recorded sessions, which lets you test a change before shipping it.
• Existing observability tools can stay as your system of record because traces can be imported.
• Apache 2.0 licensing for Kitaru and open source for ZenML allow self-hosting in your own VPC.

ZenML features

• Python pipeline and flow orchestration
• Artifact management and a basic model registry
• Model Control Plane and Artifact Control Plane on paid tiers
• Snapshots and Codespaces remote IDE on ZenML Scale
• Kitaru agent recording and replay-based evals in Python and TypeScript
• Replay on your own workers using your own model keys
• Trace import from Langfuse, LangSmith, Braintrust, Logfire and Arize Phoenix
• Run templates, schedules and API or webhook triggers
• SSO, RBAC and audit logs on enterprise tiers
• SOC 2 Type II and ISO 27001 compliance listed for ZenML Pro

ZenML integrations

• Kubernetes
• Vertex AI
• SageMaker
• AzureML
• Langfuse
• LangSmith
• Braintrust
• Logfire
• Arize Phoenix

How to get started with ZenML

1. Install the open-source package and write your workflow as Python pipelines.
2. Choose a stack, such as Kubernetes, Vertex, SageMaker or AzureML, to run the pipelines on.
3. Track artifacts, models and runs in the dashboard as pipelines execute.
4. For agents, record runs with Kitaru and replay them with your own workers and model keys.
5. Move to a hosted plan for a managed control plane, team roles and governance when you need them.

ZenML limitations

• There is no free hosted tier; the free option is the self-hosted open-source version
• Kitaru Cloud is limited to 3 agents, 2 seats and 90-day session retention
• Community support only on Open Source and Kitaru Cloud
• SSO and audit logs require an Enterprise plan

ZenML pricing

ZenML is open source, so you can use and self-host it for free. Paid plans start at $39 per month, according to the official pricing page on 24 Oct 2026.

ZenML plans

• Open Source

  Free

  self-hosted

  • Pipeline and flow orchestration
  • Artifact management
  • Basic model registry
  • Community support

• Kitaru Cloud

  $39/month

  14-day full-access trial

  • 3 agents, 2 seats
  • 90-day session retention
  • Hosted dashboard and control plane
  • Replays and experiment runs included

• ZenML Scale

  $999/month

  per workspace, by monthly executions

  • 2,000 executions, 3 projects, 5 snapshots
  • Model Control Plane
  • Artifact Control Plane
  • Codespaces remote IDE

• Enterprise

  Custom

  • SSO (SAML / OIDC)
  • Audit logs
  • RBAC (custom roles)
  • Air-gapped deployment
  • Dedicated support with SLA

Plans as listed on the official pricing page on 24 Oct 2026. Prices can change and may vary by country or exclude tax. Check the official pricing page.

ZenML alternatives

The main ZenML alternatives listed on Debuly are Metaflow, Kubeflow, MLflow, Strands Agents, Letta and OpenLIT. The table compares pricing and platforms; order is A–Z.

1. 

   Kubeflow

   Open source, Kubernetes-native AI platform made of modular projects for training, tuning, pipelines, notebooks and model management.

   Developer platformsAutomation and agents
   Open sourceVisit Kubeflow (official site)
2. 

   Letta

   Platform, formerly MemGPT, for building stateful AI agents with persistent memory, from a terminal app, SDK or API.

   Developer platformsAutomation and agentsCoding
   FreemiumVisit Letta (official site)
3. 

   Metaflow

   Open-source Python library for building ML, AI and data science workflows locally and running them at scale in your cloud.

   Developer platformsAutomation and agentsCoding
   Open sourceVisit Metaflow (official site)
4. 

   MLflow

   Open-source AI engineering platform for tracing, evaluating and deploying LLM agents, apps and classic ML models.

   Developer platformsAutomation and agentsCoding
   Open sourceVisit MLflow (official site)
5. 

   OpenLIT

   Open-source, OpenTelemetry-native platform for tracing, evaluating and guarding AI agents, with cost and GPU monitoring

   Developer platformsCodingAutomation and agents
   Open sourceVisit OpenLIT (official site)
6. 

   Strands Agents

   Free Apache-2.0 SDK for creating and running Python and TypeScript AI agents, with built-in tools, MCP, memory and multi-agent patterns.

   Developer platformsAutomation and agentsCoding
   Open sourceVisit Strands Agents (official site)

ZenML FAQ

Sources & verification

Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.

Image credits

• Logo: from zenml.io · fetched 24 Oct 2026 · Vendor trademark, used only to identify the product; removed on request
• Cover: none collected yet, so a generated tile is shown.

Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.