IBM watsonx.ai provides the full set of capabilities in the Model Hub capability group. watsonx.ai is available as both a cloud-based service and as an on-premise deployment package; enabling a hybrid deployment approach that combines on-premise and cloud-based capabilities; providing architects with the flexibility they need too meet the functional and non-functional requirements of their generative AI solutions without having to make substantial architectural trade-offs.

IBM watsonx.ai provides the Model Lifecycle Management, Model Inferencing, and Model Access Policy Management capabilities of the Model Hosting capability group. For Model Inferencing, watsonx.ai provides enterprises with the ability to deploy generative AI models as REST services using a common API. This makes it easy for generative AI solution developers to incorporate and access models within their applications, while also enabling data scientists and AI engineers to update models over time without impacting the consuming applications.

watsonx.ai offers deployment spaces to address Model Access Policy Management capabilities. Deployment spaces are access controlled collections of deployable models, data, and environments that enterprises can use to manage their generative AI models and control access to those assets.

For Model Lifecycle Management, watsonx.ai gives enterprises the ability to deploy, update, and retire / delete models over time.

IBM watsonx.ai provides the Model Fine-tuning and Embeddings Generation capabilities within the Model Customization group. Model Tuning is supported by a companion solution, Red Hat Openshift AI.

BM watsonx.governance provides the majority of the capabilities in the Model and Data Governance group. Specifically, watsonx.governance provides Model and Data Card Management, Model Catalogue Management, Model Risk Governance, and Legal and Compliance Management. Data Catalog Management is provided by IBM watsonx.data.

Model monitoring is the operational aspect of Model and Data Governance, and so IBM watsonx.governance provides many of the capabilities with the Model Monitoring group to complement the Model and Data Governance capabilities it provides. Specifically, watsonx.governance provides the HAP Detection, Model Drift Detection, Model Feedback and Improvement, Explainability, and Model Evaluation capabilities within this group.

IBM watsonx.ai provides the Prompt Engineering and Prompt Tuning capabilities within the GenAI Engineering group. Model Tuning is supported by a companion solution, Red Hat Openshift AI.

IBM watsonx Orchestrate uses natural language processing to draw from a catalog of basic and advanced skills to execute on your requests—in context and in the right order.