live wire
▸AI · Red Hat documents usage-based admission fair sharing for Kueue 1.4 on OpenShiftRed Hat Developer▸AI: Red Hat maps governed firewall changes from ServiceNow through Ansible and two human approval gatesRed Hat Developer▸CLUSTER MGMT · ACM 2.17 makes Submariner 0.24 GA with Important-rated fixesRed Hat Errata▸PLATFORM · Red Hat makes on-premises Lightspeed recommendations GA for Satellite 6.18Red Hat Errata▸SECURITY · Red Hat Hardened Images updates Tomcat 10 for nine authentication, access-control and DoS flawsRed Hat Errata▸AI · Open Data Hub 3.6.0 EA1 bundles Trainer, MLflow and llm-d componentsOpen Data Hub▸AI · Speculators 0.6.0 adds P-EAGLE parallel drafting for vLLM speculative decodingRed Hat Developer▸SECURITY · OpenShift 4.17.57 fixes seven Go and TLS CVEs in an Important-rated updateRed Hat Errata▸AI · Red Hat benchmarks local LLM guardrails with EvalHub, exposing regex accuracy and latency trade-offsRed Hat Developer▸AI · Red Hat maps silent tool-call failures across agentic pipelinesRed Hat▸API · Kuadrant 1.5.3 adds GRPCRoute policies and developer-portal API-key workflowsKuadrant▸AI · (Aug 25) IBM releases Apache-2.0 Granite 4.2 reasoning models in 3B, 8B and 30B sizesIBM Research▸JAVA · Red Hat build of Quarkus 3.33.3.SP1 fixes 13 CVEs in an Important-rated updateRed Hat errata▸AI · vLLM moves Kimi K2 RL weight sync across 384 H100s in 7.53 seconds (Aug 22)vLLM▸AI · Red Hat documents usage-based admission fair sharing for Kueue 1.4 on OpenShiftRed Hat Developer▸AI: Red Hat maps governed firewall changes from ServiceNow through Ansible and two human approval gatesRed Hat Developer▸CLUSTER MGMT · ACM 2.17 makes Submariner 0.24 GA with Important-rated fixesRed Hat Errata▸PLATFORM · Red Hat makes on-premises Lightspeed recommendations GA for Satellite 6.18Red Hat Errata▸SECURITY · Red Hat Hardened Images updates Tomcat 10 for nine authentication, access-control and DoS flawsRed Hat Errata▸AI · Open Data Hub 3.6.0 EA1 bundles Trainer, MLflow and llm-d componentsOpen Data Hub▸AI · Speculators 0.6.0 adds P-EAGLE parallel drafting for vLLM speculative decodingRed Hat Developer▸SECURITY · OpenShift 4.17.57 fixes seven Go and TLS CVEs in an Important-rated updateRed Hat Errata▸AI · Red Hat benchmarks local LLM guardrails with EvalHub, exposing regex accuracy and latency trade-offsRed Hat Developer▸AI · Red Hat maps silent tool-call failures across agentic pipelinesRed Hat▸API · Kuadrant 1.5.3 adds GRPCRoute policies and developer-portal API-key workflowsKuadrant▸AI · (Aug 25) IBM releases Apache-2.0 Granite 4.2 reasoning models in 3B, 8B and 30B sizesIBM Research▸JAVA · Red Hat build of Quarkus 3.33.3.SP1 fixes 13 CVEs in an Important-rated updateRed Hat errata▸AI · vLLM moves Kimi K2 RL weight sync across 384 H100s in 7.53 seconds (Aug 22)vLLM
upstreambeat.ai
releaseAI

Open Data Hub 3.5 connects AI Gateway, MCP lifecycle and agent deployment workflows

The release adds operator modules for gateway and MCP infrastructure while expanding the dashboard’s agent, NVIDIA NIM and NeMo Guardrails paths.

By The News Desk· Aug 17, 2026

What changed

Open Data Hub 3.5 expands the project’s generative-AI control plane with new operator modules for AI Gateway and Model Context Protocol lifecycle management, alongside dashboard work for agent deployments, NVIDIA NIM model serving and NeMo Guardrails.

At the operator layer, the 3.5 release integrates the AI Gateway and MCP Lifecycle Operator into the Open Data Hub manifests and reports their module status through the platform’s DataScienceCluster configuration. Models as a Service is nested under the AI Gateway module, tightening the relationship between shared model access and the gateway that brokers it. The release also adds ordering for provisioning and upgrades, forwards shared settings such as cluster TLS profiles and monitoring namespaces, and includes network-policy fixes for OpenShift 4.22 and disconnected environments.

The dashboard gives those services a more coherent user path. MCP deployments move into Open Data Hub namespaces with project synchronization and a condition-based status model. The release also introduces agent-operations module wiring and list/detail routes, laying the dashboard foundation for managing agent deployments as first-class AI assets.

Model-serving workflows now include NVIDIA NIM-specific options: an image selector in the deployment wizard, project API-key settings, persistent-volume caching, start and stop controls, and integration with the unified deployments table. The dashboard keeps NIM-owned inference services out of the ordinary KServe listing to avoid showing one deployment twice.

For safety controls, the generative-AI interface replaces its earlier guardrail path with NeMo Guardrails integration. The release adds initialization and status endpoints, playground support, inline guardrail configuration, per-request model discovery and generated “View Code” output.

Who is affected

The changes matter to platform teams assembling an internal AI service rather than deploying one model at a time. AI Gateway and Models as a Service affect operators providing shared model endpoints and subscription policies. MCP lifecycle management affects teams that need to deploy tool servers, associate them with projects and expose their status to application developers.

AI engineers gain dashboard paths for NIM-backed models, agent assets and guardrail-enabled experimentation. Cluster administrators inherit more components, credentials and lifecycle dependencies to govern. The modular architecture reduces in-tree coupling, but it also makes module management state, upgrade ordering, RBAC and network reachability part of the operational contract.

What to do

Before upgrading, administrators should review the 3.5 DataScienceCluster sample and decide which AI Gateway, Models as a Service and MCP lifecycle modules they intend to manage. Validate upgrade ordering and module status in a non-production cluster, particularly on OpenShift 4.22 or in disconnected installations where gateway proxy images and egress rules have specific fixes.

For NIM, confirm image access, NVIDIA credentials, storage for model caching and the distinction between NIM and KServe-owned deployments. For MCP servers, test namespace selection, project synchronization, authorization and condition-based status reporting with representative users.

Treat dashboard feature flags as deployment boundaries, not cosmetic switches. Enable agent operations, NIM workflows and NeMo Guardrails only after their backend services are ready. Finally, run guardrail tests against the same models and configurations used in production, inspect the generated deployment code, and retain the resulting settings in version control. Open Data Hub 3.5 assembles the pieces of an agent platform; operators still need to define who can publish models, attach tools and change runtime safety controls.

sources

Filed by The News Desk. Corrections: desk@upstreambeat.ai · Our standards →

comments · 0

    Comments are moderated before they appear. Your email is used once to confirm it is you — never shown, never sold. Corrections and questions get an answer from the desk when we have one.

    more from the desk