Kubeflow
Free tierThe open-source foundation of tools for AI Platforms on Kubernetes
Free·Technical·API available·Open source
Key strengths
CNCF graduated projectKubernetes-native AI/ML platformModular and composable subprojectsLarge community with 3K+ contributorsSupports distributed training, AutoML, pipelines, and model registry258M+ PyPI downloads
Completely free
Founded 2018
Self-hostable
Facts verified against kubeflow.org on
Get alerts when Kubeflow changes
Free. We'll email you when pricing, the free tier, API access or open-source status changes.
No ratings yet
Kubeflow — Documentation & Developer Resources
Kubeflow is self-hostable and deploys on any Kubernetes-conformant cluster. Its modular architecture means each subproject (Training Operator, Pipelines, Katib, Model Registry, etc.) can be installed and configured independently.
Integration surface:
- Training frameworks: PyTorch, JAX, XGBoost, DeepSpeed, Megatron, MLX
- Data & model ecosystems: HuggingFace, Apache Spark
- Infrastructure: Kubernetes (CRDs, operators, RBAC)
Architecture highlights:
- Workloads are defined as Kubernetes Custom Resources — enabling declarative, GitOps-compatible ML infrastructure
- Distributed training jobs are managed via the Training Operator, supporting multi-node, multi-GPU topologies
- Pipelines are expressed as DAGs and executed natively on Kubernetes
- AutoML (Katib) runs hyperparameter search and neural architecture search as Kubernetes jobs
- Model Registry provides versioned lineage tracking for trained models
Community & ecosystem:
- CNCF graduated project (highest maturity tier)
- 33,100+ GitHub stars, 3,000+ contributors, 258M+ PyPI downloads
- Founded 2018; long-term, stable open-source governance
Alternatives
See all alternatives Alternatives to Kubeflow
Is Kubeflow yours? Claim the listing & get the badge