A new System One endpoint serves six decision-model sizes for typed text, JSON, image and video questions, but the code is still nightly-only.

Red Hat has published updated images, but the linked 3.4 release notes still describe 3.4.4 and the advisory says upgrade documentation will be refreshed shortly.

The Q3 Security Signals report treats vulnerabilities, configurations, credentials and sensitive resources as one attack path—shifting priority from isolated scores to reachable outcomes.

An open pull request reports as much as 1.92× faster non-speculative decode on GB300 while narrowing where agent-assisted kernel code can run.

The first 1.x release improves coordinated rollout and recovery, but operators must review changed defaults and a removed compatibility path before upgrading.

CVE-2026-99999 is rated Important by Red Hat, but the vendor’s active-exploit designation and lack of a listed fix make it an immediate OpenShift watch item.

The release drops Python 3.10, restricts browser origins by default and bundles three named security fixes alongside serving and hardware updates.

A working Muse Spark Open stack shows what operators can use now; broader day-zero model coverage and performance publication remain commitments.

The new recipe preserves the September model’s deployment shape across AMD and NVIDIA hardware, but its published validation still belongs to the older weights.

Red Hat’s three-month notice gives remaining 4.12 operators a firm deadline to complete their move to a supported OpenShift release.

A new engineering guide traces false-positive and flaky tests to route structure, serialization, timing and infrastructure startup—not just test code.

Three new recipes turn model support into reproducible deployment contracts across AMD text inference, single-GPU speech and NVIDIA image generation.

A Red Hat engineering walkthrough connects policy gates, an SRE agent and human approval into a repeatable failed-build response pattern.

The implementation supports tool calling and structured output, while vLLM reports a 2.71× batch-one decode speedup from seven model-specific changes.

A day-zero image packages text-to-image and multi-reference editing behind OpenAI-compatible endpoints, but the checkpoint is gated and the benchmark is one GB200 run.

The verified recipe separates model-weight fit from KV-cache capacity, with distinct tensor-, expert- and data-parallel choices for B300 and H200 systems.

The guidance connects admission-webhook conflicts, rootless Podman SCCs, GitOps configuration, capacity monitoring and five disconnected dependencies.

Connectivity Link 1.5 adds tenant isolation, protocol negotiation, progressive tool discovery, audit telemetry and NeMo guardrails for agent tool traffic.

The new LTS line adds preview semantic routing, fail-closed policy components and a broader MCP and observability surface for agentic integrations.

The initial integration spans NVIDIA and AMD deployments, a DSpark draft model and limited TPU access, while the main code change still awaits merge.

The result is a software-stack comparison, not a clean GPU shootout, but the engineering behind it offers a useful optimization map.

The hands-on workflow takes teams from a containerized OpenAI-compatible endpoint through concurrency tests and a local chat interface before they scale out.

A merged model-integration patch and six new recipes put typed multimodal decisions behind vLLM’s System One API, although users need a nightly build for now.

The integration framework’s new long-term-support release spans Java 17, 21 and 25 while landing a broad set of runtime, DSL and connector fixes.

Red Hat’s architecture guide explains why higher VM density depends on coordinated reclaim, tiering, pressure signals and cluster rebalancing—not one overcommit ratio.

New Red Hat AI checkpoints split the performance problem between lower-precision execution and parallel speculative decoding on NVIDIA Hopper GPUs.

A directional five-year model argues that tiering enterprise AI traffic across open-weight and frontier models can cut TCO, but the result depends heavily on workload mix and utilization.

The Q3 report combines Lightwell package observations with an explicitly modeled Java scenario, estimating how shared remediation could avoid duplicated work.

The design separates the delivery of signed fixes for pinned dependencies from the approvals, testing, rollback and audit work needed to deploy them.

An online-fitted estimator lets more speculative-decoding setups trim low-value draft tokens without retraining existing checkpoints.

The new guidance separates fixed-capacity deployments from demand-driven scaling and points operators toward service-level inference metrics rather than GPU utilization.

The quarterly engineering update emphasizes safer CI, more predictable device-state handling and fixes for large configurations and connection timeouts.

The guidance covers centralized SCM authentication, private extension registries, certificate trust and restricted-network IDE workflows.

A new engineering comparison says benchmark parity makes traffic shape, hardware support and ecosystem fit the deciding factors.

The Kafka operator release adds Kafka 4.2.2 and early cert-manager integration while removing Kubernetes 1.30 and 1.31 from support.

A 52× HLS speedup for about $26 shows why cache-aware serving may matter as much as model choice in automated design exploration.

A concrete device-plugin pattern turns mainframe cryptographic hardware into a schedulable OpenShift resource, with explicit limits around host setup and pod privilege.

The unsupported preview also advances the inference stack, adds a separate vLLM-Omni technology-preview image and narrows the supported upgrade path.

Red Hat’s walkthrough uses the same bootc image for disposable checks, persistent libvirt development and qcow2 disk output.

A two-shard tech-preview configuration sharply reduced slow applies during event storms, but added memory cost and regressed under one quota-pressure workload.

The Kubernetes deployment adds API-key authentication, subscription-aware access and token quotas in front of vLLM and KServe model endpoints.

Ferrocene 26.05 supports a qualified RHIVOS 2 target now, with the full joint solution planned for the second half of 2027.

The project’s new recipes belong after dependency and template updates, while larger migrations such as MUI-to-BUI remain deliberate, review-heavy work.

A new worked example uses Camel’s AI Tool component and built-in MCP server to expose three narrowly defined CRM operations without handing the model arbitrary SQL access.

A new IBM Research explainer connects llm-d’s router, cache tiers and split worker pools to the uneven traffic generated by AI agents.

The PeopleMesh quickstart combines Docling, Granite, pgvector and Keycloak, while exposing the design choices behind structured extraction and semantic reranking.

The field report connects OpenShift failure injection to release gates, virtualization migrations, database recovery and payment SLOs.

Live Red Hat documentation confirms a production JobSet operator, a preview cluster-diagnostics MCP server and security changes that matter to multitenant operators.

Decision 3.0 ships six Apache-2.0 models that answer choice, yes/no and scoring questions with probabilities instead of generated prose.

The first Quarkus 4 beta raises the Java baseline, refreshes the networking and persistence stack, and asks extension maintainers to move before a planned late-November final release.