IBM watsonx.ai v2.4 Enhances Governed AI Development for Enterprises

IBM has announced the release of watsonx.ai v2.4, an update aimed at transforming how enterprises build, deploy, and govern artificial intelligence solutions. The new version emphasizes greater model choice, enhanced governance frameworks, modernized runtimes, and increased adaptability for hybrid cloud infrastructures. This strategic update is designed to help organizations move their AI projects from experimentation to production with greater confidence and control. A core component of the v2.4 release is the expanded Model Gateway. This feature addresses the growing challenge of managing diverse foundation models from various providers and hosting environments. By offering a unified, OpenAI-compatible API interface, the Model Gateway simplifies model access for developers while providing administrators with consistent controls for access, policy enforcement, usage monitoring, and auditing. Key enhancements to the Model Gateway include an administrative user interface for configuring providers, models, secrets, and policies, as well as integration with Prompt Lab and support for locally hosted models. The update also broadens support for external model providers like Mistral, Cohere, and xAI, and introduces user account-based authentication for enhanced security. Beyond model access and governance, watsonx.ai v2.4 brings substantial improvements to data preparation and management tools. The Synthetic Data Generator now features an API for programmatic use of structured synthetic data generation, allowing users to integrate this capability directly into their existing workflows. It also adds Large Language Model (LLM) selection for unstructured synthetic data generation. Furthermore, the Data Refinery component has been upgraded with support for Spark 3.5, Microsoft Azure Databricks, and Vertica, alongside new functionalities like job parameterization, the ability to cancel starting jobs, and Review Folders support. These improvements are critical for data engineers and scientists seeking to efficiently prepare and manage data across complex hybrid and multi-cloud ecosystems. IBM also highlighted runtime modernization efforts within watsonx.ai v2.4, advising clients to review and plan for necessary migrations. Specifically, users are encouraged to transition their notebooks, custom environments, and deployments from IBM Runtime 24.1 to IBM Runtime 25.1 to avoid service disruption, as support for the older runtime is slated for removal in early July 2026. This focus on continuous modernization ensures that the platform remains robust and capable of supporting emerging AI workloads.