Tools

Workflow Orchestration Component

CI / CD component

ML Metadata Stores / Experiment Tracking

Model Training Infra

Model Registry

ML Testing

Model Serving Component

Monitoring Component

Is Open Source

On prem

Cloud

VPC

User Interface

Community Support / Documentation

Security and Governance

Integrations with other tools/ platforms

LLM support

Notes

Links

Apache Airflow

Y

Y

Y

Y

Y

Y

Widely used orchestration tool by the community

AWS (SageMaker + additional AWS services)

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Uses MLFlow for experiment tracking

Azure Databricks

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Databricks platform can be used in AWS, Azure or GCP

https://azure.microsoft.com/en-gb/products/databricks#layout-container-uidfe1a

Azure (Machine Learning + additional Azure services)

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Uses MLFlow for experiment tracking

BentoML

Y

Y

Y

Y

Y

Y

Y

BentoCloud provides a fully managed platform, freeing you from infrastructure concerns and allowing you to focus on shipping AI applications. BentoCloud is built for multi-cloud environments, supporting AWS, Azure, and GCP (Google Cloud Platform). We also provide private deployment for enterprise customers who need advanced security control and specialized AI hardware support.

https://bentoml.com/cloud https://github.com/bentoml/BentoML https://github.com/bentoml/OpenLLM

Censius AI

Y

Y

Y

Y

Provides AI observability which is Monitoring+Accountability+Explainability

https://mlops.community/learn/monitoring/censius/ https://censius.ai/ https://censius.ai/mlops

Comet ML

Y

Y

Y

Y

Y

Y

https://www.comet.com/site/products/llmops/

DagsHub

Y (see notes)

Y (see notes)

Y

Y

Y

Y

Y

Y

Uses MLFLow's Experiment Tracking and Model Registry

https://dagshub.com/product/

Data Version Control (DVC)

Y (see notes)

Y (see notes)

Y

Y

Y

Uses Git under the hood for Experiment Tracking and Model Registry

https://iterative.ai/ https://iterative.ai/blog/introducing-dvc-studio https://dvc.org/doc/studio

DataRobot (Algorithmia)

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

https://www.datarobot.com/platform/mlops/

Evidently

Y

Y

Y

Y

Y

Y

Y

https://www.evidentlyai.com/

Fiddler AI

Y

Y

Y

Y

Y

Y

Y

Y

Provides AI observability which is Monitoring+Accountability+Explainability

https://www.fiddler.ai/mlops https://www.fiddler.ai/llmops

Flyte

Y

Y

Y

Y

Y

workflow automation tool built on top of Kubernetes

https://flyte.org/blog/fine-tuning-large-language-models-with-declarative-ml-orchestration

GCP (VertexAI + additional GCP services)

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Hydrosphere

Y

Y

Y

Y

Y

The platform helps serving and monitoring on top of kubernetes.

https://docs.hydrosphere.io/

Kedro

Y

Y

Y

Y

Y

Kedro is an open-source Python framework hosted by the Linux Foundation (LF AI & Data). Kedro uses software engineering best practices to help you build production-ready data science code. Check FAQs

https://kedro.org/#features https://docs.kedro.org/en/0.17.1/index.html https://github.com/kedro-org/kedro-viz https://medium.com/@getindatatechteam/running-kedro-everywhere-machine-learning-pipelines-on-kubeflow-vertex-ai-azure-and-airflow-fb7e834e6b6e

Kubeflow

Y

Y

Y

Y

Y

Y

Y

Y

Y

Uses KServe for Model Serving

Metaflow

Y

Y

Y

Y

Y

Y

Y

https://docs.metaflow.org/getting-started/infrastructure

MLFlow

Y

Y

Y

Y

Y

Y

Y

Y

Y

Cloud, On prem and Hybrid

Neptune

Y

Y

Y

Y

Y

Y

Y

Y

https://docs.neptune.ai/integrations/

Pachyderm

Y

Y

Y

Y

Y

Y

Y

The Community Edition is Open Source

Polyaxon

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

The Community Edition is Open Source Offers distributed training

https://polyaxon.com/features/ https://polyaxon.com/product/ https://polyaxon.com/docs/management/ https://github.com/polyaxon/polyaxon

Prefect

Y

Y

Y

Y

Y

Y

Y

Y

Y

Paid pricing for startups to large orgs Cloud Convenience; On-Prem Security. Prefect's hybrid execution model keeps your code and data private while still taking full advantage of our managed workflow orchestration service. It's so innovative, we patented it. Execution in your cloud; orchestration in ours. We designed the hybrid model to meet the strict standards of major financial institutions and companies that work with regulated data. Prefect Cloud never receives your code or data. It orchestrates Prefect 1.0, running on your private infrastructure, by exchanging state information and metadata. Lang chain prefect

https://github.com/PrefectHQ/prefect https://www.prefect.io/why-prefect/hybrid-model/ https://www.prefect.io/guide/blog/keeping-your-eyes-on-your-ai-tools/ https://github.com/PrefectHQ/langchain-prefect

Seldon

Y

Y

Y

Y

Y

Y

Y

Y

Paid pricing for startups to large orgs Explainable AI

https://github.com/SeldonIO/seldon-core https://www.seldon.io/explain-with-seldon https://www.seldon.io/manage-with-seldon

Valohai

Y

Y

Y

Y

Y

Y

Y

Y

https://valohai.com/blog/the-mlops-stack/

Weights & Biases

Y

Y

Y

Y

Y

Y

Y

https://wandb.ai/site/mlops-maturity-assessment https://wandb.ai/wandb_fc/mlops_course/reports/Using-W-B-Model-Registry-to-Manage-Models-in-Your-Organization--VmlldzozMjI1NTc2#:~:text=In%20this%20video%20from%20our,their%20metadata%2C%20metrics%20and%20lineage. https://wandb.ai/site/solutions/deployment-options

ZenML

Y

Y

Y

Y

Y

Y

Y

Y

Y

Paid pricing for startups to large orgs It is a pipeline orchestrator ZenML Cloud is built on top of the core ZenML open-source offering. Your worklow won't change - your ML workloads will still run on your stack on your infrastructure. However, your ZenML server will be managed by us on our cloud. Compliance and governance won’t change - you are still in charge of where the data is stored, where processing happens and where models get deployed. ZenML runs out-of-the-box with powerful orchestrators like Kubeflow and Airflow, and if the tools you want to use do not have a ZenML integration, you can easily integrate them yourself.

https://docs.zenml.io/platform-guide/set-up-your-mlops-platform https://www.zenml.io/cloud https://github.com/zenml-io/zenml https://zenml.io/integrations

MLRun

Y

Y

Y

Y

Y

Y

Y

Deepchecks

Y

Y

Y

Y

Y

Y

Open source Testing Paid pricing for usage

Ray IO

Y

Y

Y

Y

Y

https://docs.ray.io/en/latest/index.html#

Wallaroo AI

Y

Y

Y

Y

Y

Y

Y

https://wallaroo.ai/ https://portal.wallaroo.community/ - community version

Anyscale

Y

Y

Aimstack

Y

Y

Y

Y

Mage AI

Y

Y

Y

Y

Y

Y

ETIQ AI

Y

Y

GitHub

Y

Y

Y

Y

Legends

Yellow

optional

Grey

Not required for maturity level 1 (i.e. Google's MLOPs Maturity level 1)

Green

Main components

Blue

Additional information