Back to Gap Map Infrastructure / ML orchestration

Kubeflow Katib

open source / Overall score: 2.4

Kubernetes-native hyperparameter tuning and neural architecture search. An Experiment declares the search space, objective and algorithm; Katib's controllers create Trials, run them in parallel up to a declared limit, collect the objective metric from each and stop the Experiment when the goal or the trial budget is reached. It is framework-agnostic and tunes whatever the trial runs.

Ships as a Kubernetes controller with no package channel, so adoption bands on stars. Verified 2026-09-15 via the repository and its recursive tree.

Openness

5 high confidence

5.0

license: Apache-2.0(OSI)
source: public(kubeflow/katib is the whole controller set)
core-gated: ungated(a 1,533-entry recursive tree carries one root LICENSE and no path matching enterprise, ee, commercial or proprietary)

Apache-2.0 under CNCF governance. The 1,533-entry recursive tree carries one governing LICENSE and no enterprise, ee or commercial path, and no vendor sells a fuller Katib.

https://github.com/kubeflow/katib recorded 2026-09-15
Repository page for kubeflow/katib: Apache-2.0 license, public and unarchived, described as 'Automated Machine Learning on Kubernetes'. Establishes the license and that the source is public.
https://api.github.com/repos/kubeflow/katib/git/trees/master?recursive=1 recorded 2026-09-15
Full untruncated recursive tree of the default branch, 1,533 entries. One root LICENSE. No path matches enterprise, ee, commercial or proprietary. A tree lists paths; it is cited for that and for nothing about the vendor's offerings.
https://www.cncf.io/projects/kubeflow/ recorded 2026-09-16
Kubeflow's project page on cncf.io, establishing CNCF hosting rather than vendor ownership, so no commercial party is positioned to withhold functionality from the published source.

Adoption

2 low confidence

2.0

1,701 GitHub stars. The project ships as container images and a Helm chart and publishes no package-registry artifact, so no usage-volume channel exists for it and the band rests on stars, which measure attention rather than use. The stars route caps at level 3 for that reason.

https://github.com/kubeflow/katib recorded 2026-09-15
Repository page showing the star count for kubeflow/katib.

Capability

3 medium confidence

3.0

Admitted on the ML-job-model limb, and on the clause of it that names a tuning trial directly. Band 3 rather than higher: Katib decides how many trials run and when they stop, and leaves where each trial's pods land to the cluster scheduler.

https://github.com/kubeflow/katib recorded 2026-09-15
Repository page and README for kubeflow/katib describing Experiments, Trials, parallel trial execution and objective-metric collection.

Verified 2026-09-15