Brian Stevens & Joe Fernandes, Red Hat | Red Hat Summit 2025
summary
Red Hat outlines an open inference stack centered on VLM open-source inference server and the Red Hat AI Inference Server that runs any model on any accelerator and any cloud; they claim 4–5x performance gains in the last year and note VLM matures in about 22 months of public development, already matching vendor stacks on Nvidia while adding AMD/Intel/TPU support. Kubernetes-based orchestration via the LMD project and Llama Stack integration enable tool calls, MCP toolchain support, and per-workload SLOs.
AI-written summary · about
