The Cloud Native Computing Foundation announced on August 17, 2026 that Kubeflow has graduated, marking the open-source Kubernetes platform as a mature option for production data processing, training, fine-tuning, and inference.

CNCF said Kubeflow has more than 6,600 contributors across more than 1,000 organizations and more than 33,000 GitHub stars across its repositories since joining as an incubating project in 2023. The foundation also cited nearly 260 million PyPI downloads for Kubeflow Python packages, with adopters including Bloomberg, NVIDIA, Red Hat, LinkedIn, and Spotify.

What graduation required

To graduate, Kubeflow completed a third-party security audit, stood up a formal steering committee, adopted the CNCF Code of Conduct, and maintains a CII Best Practices Badge. CNCF CTO Chris Aniszczyk called the milestone confirmation that Kubeflow is a mature option for enterprise AI workloads on Kubernetes.

The project began at Google in 2017 and now positions itself as a vendor-neutral stack spanning interactive development through model serving. CNCF said the near-term roadmap emphasizes LLM orchestration, post-training and fine-tuning, large-scale data engineering, and agentic Data and AI workloads.

Decoded Take

Graduation is less about new features and more about procurement signal. “CNCF Graduated” clears risk and vendor-selection conversations that incubating projects often stall. As enterprises push agentic and LLM workloads onto Kubernetes, portable MLOps foundations matter more than another proprietary control plane. Watch whether Kubeflow’s LLM and agentic roadmap keeps pace with managed platforms, or whether graduation mainly accelerates adoption of the training and serving pieces teams already trust.