Skip to main content

Deployment Guides

These guides mirror the well-lit path deployment recipes in the llm-d repository — the manifests and step-by-step instructions referenced by the "Deploy" section of each well-lit path.

A guide isn't a single flat thing: it varies by model, model server, accelerator, and router mode (standalone vs. gateway), with observability layered on top. Each page below documents those variations for its guide; browse well-lit paths first to decide which capability or workload you need, then come here for the manifests.

Guides are migrated onto the site incrementally — a guide not yet listed in the sidebar still resolves from its well-lit path's "Deploy" link, which falls back to the source recipe on GitHub.