Red Hat OpenShift AI: API Tiers
For Red Hat OpenShift AI, APIs are stable within a major release for a minimum of 9 months from the announcement of deprecation. In addition, the following statements apply:
-
APIs may be deprecated and removed within a major release.
-
APIs may have a longer deprecation window than the minimum.
All commercially-supported APIs, components, and features are associated under one of the following support levels:
API Tier 1
APIs in Tier 1 are stable within a major release for a minimum of 18 months.
API Tier 2
APIs in Tier 2 are stable within a major release for a minimum of 9 months.
API Tier 4
No compatibility is provided for APIs in Tier 4. A Tier 4 API can change at any point. The capabilities of Tier 4 APIs are not intended for use outside of OpenShift AI components.
API Tier Beta
APIs in Tier Beta are for Technology Preview features. Red Hat provides at least a manual migration path for any breaking changes.
API Tier Alpha
APIs in Tier Alpha are for unsupported Developer Previews. Red Hat does not provide backward compatibility guarantees.
The following table lists the API tier for each OpenShift AI API version example (customer accessible endpoint).
| API version example | API tier |
|---|---|
hardwareprofiles.infrastructure.opendatahub.io/v1 | Tier 1 |
notebooks.kubeflow.org/v1 | Tier 1 |
odhapplications.dashboard.opendatahub.io/v1 | Tier 1 |
odhdocuments.dashboard.opendatahub.io/v1 | Tier 1 |
odhquickstarts.console.openshift.io/v1 | Tier 1 |
| OpenVino Model Server API (Open Inference Protocol) v2.0 | Tier 1 |
pytorchjobs.kubeflow.org/v1 | Tier 1 |
rayclusters.ray.io/v1 | Tier 1 |
| CodeFlare SDK 0.16.0 | Tier 2 |
datasciencecluster.opendatahub.io/v1 | Tier 2 |
dscinitialization.opendatahub.io/v1 | Tier 2 |
| DSP Server API (REST endpoint) v2 | Tier 2 |
inferenceservice.serving.kserve.io/v1beta1 | Tier 2 |
llminferenceservices.serving.kserve.io/v1alpha1 | Tier 2 |
llminferenceserviceconfigs.serving.kserve.io/v1alpha1 | Tier 2 |
| Model Registry Python client 0.3.3 | Tier 2 |
| Model Registry REST API for Dashboard v1alpha3 | Tier 2 |
modelregistry.opendatahub.io/v1beta1 | Tier 2 |
rayjobs.ray.io/v1 | Tier 2 |
servingruntime.serving.kserve.io/v1alpha1 | Tier 2 |
| TrustyAI Python library 0.6.0 | Tier 2 |
| TrustyAI REST API v1alpha1 | Tier 2 |
trustyai.opendatahub.io/v1alpha1 | Tier 2 |
*.argoproj.io/v1alpha1 | Tier 4 |
codeflares.components.platform.opendatahub.io/v1alpha1 | Tier 4 |
| Dashboard REST API | Tier 4 |
dashboards.components.platform.opendatahub.io/v1alpha1 | Tier 4 |
| Pipelines REST API v2beta1 | Tier 4 |
datasciencepipelinesapplications.opendatahub.io/v1 | Tier 4 |
feastoperators.components.platform.opendatahub.io/v1alpha1 | Tier 4 |
featuretrackers.dscinitialization.opendatahub.io | Tier 4 |
kserves.components.platform.opendatahub.io/v1alpha1 | Tier 4 |
kueues.components.platform.opendatahub.io/v1alpha1 | Tier 4 |
modelcontrollers.components.platform.opendatahub.io/v1alpha1 | Tier 4 |
modelmeshservings.components.platform.opendatahub.io/v1alpha1 | Tier 4 |
modelregistries.components.platform.opendatahub.io/v1alpha1 | Tier 4 |
monitorings.services.platform.opendatahub.io/v1alpha1 | Tier 4 |
| Ray cluster job API (REST endpoint) 4.0.0 | Tier 4 |
rays.components.platform.opendatahub.io/v1alpha1 | Tier 4 |
trainingoperators.components.platform.opendatahub.io/v1alpha1 | Tier 4 |
trustyais.components.platform.opendatahub.io/v1alpha1 | Tier 4 |
workbenches.components.platform.opendatahub.io/v1alpha1 | Tier 4 |
llamastackdistributions.llamastack.io/v1alpha1 | Beta |
*.serving.kserve.io/v1alpha1[1] | Alpha |
auths.services.platform.opendatahub.io/v1alpha1 | Alpha |
odhdashboardconfigs.opendatahub.io/v1alpha | Alpha |
| Llama Stack REST API | Alpha |
Footnotes
- All APIs in
*.serving.kserve.io/v1alpha1are currently in Alpha, except forinferenceservice,llminferenceserviceconfigs,llminferenceservices, andservingruntime.