As enterprises accelerate their generative AI ambitions, the need for a unified, reliable, and secure AI inferencing stack has never been greater. IBM AI Optimizer for IBM Z and IBM LinuxONE version 3.1.0 is that stack — purpose-built for organizations running mission-critical workloads on IBM Z and IBM LinuxONE. Planned for availability on 30 April 2026, this release brings together model onboarding, serving, and monitoring into a single integrated solution, while expanding the program's scope to explicitly support both IBM Z and IBM LinuxONE environments.
What's new in IBM AI Optimizer for IBM Z and IBM LinuxONE
Version 3.1.0 introduces a set of capabilities designed to simplify deployment, improve visibility, and give enterprises greater control over their AI inferencing workloads:
-
Simplified deployment and administration: Delivered as a single Logical Partition (LPAR) image, the solution bundles the operating system, curated AI models, container runtime, observability tooling, and management UI into one pre-configured package. Administrators can get the entire inferencing stack up and running through a single installation process, minimizing setup complexity and accelerating time to production.
-
Real-time monitoring and observability: Advanced real-time monitoring through the use of Prometheus (backend) and Grafana (visualization) surfaces key metrics including inferencing performance, resource utilization, LLM usage, and aggregated cross-application monitoring. Integration with the OpenTelemetry collector enables seamless telemetry ingestion and unified observability across hybrid environments.
-
Inferencing optimization for LLMs running on Spyre: LLMs running on the IBM Spyre Accelerator are automatically detected and registered for inferencing optimization. Users can configure custom routing plans and group multiple LLMs using customizable tags, aligned to OpenAI API standards, for greater flexibility and control.
-
External LLM registration: External LLMs can be registered and brought into the unified inferencing environment alongside local models on the IBM Spyre Accelerator. Once registered, they can be tagged, grouped, and optionally monitored through the cross-platform observability dashboard, providing a complete view of GenAI activity across all models.
Use case scenarios
AI Optimizer is purpose-built for enterprises running mission-critical workloads on IBM Z and IBM LinuxONE. The following scenarios represent the highest-impact deployments across regulated industries:
-
Fraud detection at transaction speed (banking & financial services): Banks processing millions of transactions daily need fraud detection that is fast, accurate, and compliant. With AI Optimizer, fraud and credit risk models run directly on the IBM Spyre Accelerator, keeping sensitive financial data on-platform, satisfying data residency regulations, and delivering inferencing at the speed transactions demand. Unified monitoring via Grafana gives operations teams real-time visibility into model performance and resource utilization across every inferencing workload.
-
Operational intelligence at scale (manufacturing & supply chain): Manufacturers running ERP and supply chain systems on IBM Z can now bring AI inferencing directly into those workflows. AI Optimizer's single LPAR (Logical Partition) deployment means teams can stand up inferencing capability without overhauling existing infrastructure, routing time-sensitive predictions locally on Spyre while tapping external LLMs for broader analytical tasks, all within a single governed framework.
-
Unified AI agent orchestration (cross-industry): Enterprises building AI agent workflows with watsonx Assistant for Z (WXA4Z) can use AI Optimizer as the inferencing backbone by routing sensitive, low-latency tasks to on-platform Granite models on Spyre, while seamlessly delegating broader reasoning to registered external LLMs. The cross-application observability dashboard gives teams a single pane of glass across all models, whether local or external.
These are just a few examples of how enterprises are putting AI Optimizer to work. The possibilities span industries and workloads, and the best way to see what it can do is to watch it in action.
See IBM AI Optimizer for IBM Z and IBM LinuxONE in action
🎥 Watch Demo:
📖 Overview of IBM AI Optimizer for IBM Z and IBM LinuxONE
If you are interested in learning more about IBM AI Optimizer for IBM Z and IBM LinuxONE, check out the announcement, or connect with a verified IBM Partner for more options.