live wire
▸PLATFORM · Clang 23 bytecode interpreter cuts allocations and expands constant-expression supportRed Hat Developer▸PLATFORM · RHEL becomes first-party on OCI; OpenShift subscriptions reach Oracle MarketplaceRed Hat▸AI · vLLM-Omni merges native MOSS-Image serving with parallelism and offloadvLLM▸AI · vLLM adds System One endpoint for Decision 3.0 multimodal routing models (Oct. 15)vLLM▸SECURITY · Red Hat rates OpenShift AI 3.4.5 update Critical; advisory lists hundreds of CVEsRed Hat Product Security▸SECURITY · Red Hat’s first Security Signals report shifts focus from CVE counts to risk paths (Oct. 15)Red Hat▸PLATFORM · IBM adds non-destructive FlexVolume-to-CSI migration tool for OpenShift object storageIBM Cloud▸vLLM tests agent-built Kimi K3 decode path at up to 1.9× speedup; PR remains openvLLM▸AI · LeaderWorkerSet 1.0 changes rollout defaults for disaggregated AI workloadsKubernetes SIGs▸SECURITY · Red Hat flags active exploitation of kernel flaw affecting OpenShift CoreOS; no fix listedRed Hat Product Security▸AI · Red Hat sets Dec. 9 showcase for AI 3.6 agent governance and model servicesRed Hat▸vLLM 0.32.0 drops Python 3.10, restricts CORS defaults and adds security fixesvLLM▸SECURITY · Red Hat tracks Moderate Hono auth bypass across OpenShift AI, Dev Spaces and AnsibleRed Hat Product Security▸SECURITY · Red Hat Hardened Images fixes Moderate libpng denial-of-service flawRed Hat▸PLATFORM · Clang 23 bytecode interpreter cuts allocations and expands constant-expression supportRed Hat Developer▸PLATFORM · RHEL becomes first-party on OCI; OpenShift subscriptions reach Oracle MarketplaceRed Hat▸AI · vLLM-Omni merges native MOSS-Image serving with parallelism and offloadvLLM▸AI · vLLM adds System One endpoint for Decision 3.0 multimodal routing models (Oct. 15)vLLM▸SECURITY · Red Hat rates OpenShift AI 3.4.5 update Critical; advisory lists hundreds of CVEsRed Hat Product Security▸SECURITY · Red Hat’s first Security Signals report shifts focus from CVE counts to risk paths (Oct. 15)Red Hat▸PLATFORM · IBM adds non-destructive FlexVolume-to-CSI migration tool for OpenShift object storageIBM Cloud▸vLLM tests agent-built Kimi K3 decode path at up to 1.9× speedup; PR remains openvLLM▸AI · LeaderWorkerSet 1.0 changes rollout defaults for disaggregated AI workloadsKubernetes SIGs▸SECURITY · Red Hat flags active exploitation of kernel flaw affecting OpenShift CoreOS; no fix listedRed Hat Product Security▸AI · Red Hat sets Dec. 9 showcase for AI 3.6 agent governance and model servicesRed Hat▸vLLM 0.32.0 drops Python 3.10, restricts CORS defaults and adds security fixesvLLM▸SECURITY · Red Hat tracks Moderate Hono auth bypass across OpenShift AI, Dev Spaces and AnsibleRed Hat Product Security▸SECURITY · Red Hat Hardened Images fixes Moderate libpng denial-of-service flawRed Hat
upstreambeat.ai
releaseAI

RHEL AI 3.6 early access moves its base to RHEL 9.8 EUS

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

RHEL AI base changes and a separate multimodal preview image.
AI-generated illustration
By The News Desk· Oct 13, 2026the quick take — two AI hosts go live when you do

Red Hat has published the first early-access release of Red Hat Enterprise Linux AI 3.6, moving the product’s operating-system base from RHEL 9.6 to RHEL 9.8 Extended Update Support and refreshing its inference runtime. Red Hat labels the build unsupported and unsuitable for production, making it a test target rather than a deployment recommendation. The 3.6 early-access release notes document the changes and limitations.

A new base and a narrower upgrade gate

The base-image change matters for teams that validate drivers, accelerators and deployment automation against a specific RHEL minor line. The same release notes say an in-place upgrade to 3.6 early access is supported from RHEL AI 3.0 or later, but not directly from RHEL AI 1.5. That leaves older installations needing an intermediate migration plan before they can evaluate the preview.

The bundled inference stack moves to vLLM 0.26.0, while the Red Hat AI Inference component is identified as 3.6.0-fast.1. Those version shifts give platform teams a concrete compatibility target for model, kernel and accelerator testing, but the early-access status means the results should not be treated as a production support commitment.

Multimodal serving arrives as a separate preview

RHEL AI 3.6 also introduces vLLM-Omni as a technology preview. Red Hat packages it in a separate CUDA image based on vLLM-Omni 0.26.1 rather than folding it into the standard inference image. That separation is operationally useful: teams can evaluate multimodal serving without quietly changing the runtime used by existing text-model deployments, while still keeping the preview outside production estates.

Another technology preview exposes Anthropic cache-token information. This can help operators inspect cache behavior and cost-related token accounting in compatible workflows, but it carries the same preview caveat.

Hardware paths remain uneven. Red Hat says the ROCm image does not include the Model Optimization Toolkit, so AMD GPU evaluators should not assume feature parity with the CUDA path when designing 3.6 tests.

What platform teams should do

The practical next step is a disposable validation environment: confirm the RHEL 9.8 EUS base against local security controls, test the supported upgrade route from a 3.x installation, and evaluate the standard and Omni images separately. Production clusters should remain on supported releases until Red Hat moves these capabilities beyond early access.

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