RHEL AI products today provides official containers and CI builds for Ampere (SM 80/86) and Hopper (SM 8*), but lacks support for earlier generations (Volta SM 70/72, Turing SM 75) and the newest Blackwell (SM 100/101/120). Customers spinning up inference workloads on those GPUs face manual unsupported NVIDIA architecture error. Extending our build pipelines and runtime logic to cover Volta, Turing, and Blackwell will deliver a seamless “vllm pull & run” experience across all mainstream NVIDIA GPUs—reducing friction, lowering support overhead, and accelerating time‑to‑impact for AI teams.