Examples
Reference deployments for SGLang, TensorRT-LLM, vLLM, and custom Dynamo backends.
Use these examples when you want a concrete starting point instead of a conceptual guide. The examples in the Dynamo repository track main; if you are using a stable release, prefer the examples from the matching release branch or the versioned recipes in these docs.
Start Here
- Hello World custom backend — a minimal GPU-unaware graph that demonstrates Dynamo’s component model.
- Kubernetes Quickstart — run a model on Kubernetes with the current recommended path.
- CLI Getting Started — run Dynamo locally from the CLI.
Backend Examples
- vLLM local deployment examples — CLI launch patterns for vLLM.
- SGLang local deployment examples — CLI launch patterns for SGLang.
- TensorRT-LLM local deployment examples — CLI launch patterns for TensorRT-LLM.
Kubernetes Templates
- vLLM DGD templates — aggregated, disaggregated, and multinode Kubernetes manifests.
- SGLang DGD templates — aggregated and disaggregated Kubernetes manifests.
- TensorRT-LLM DGD templates — TensorRT-LLM Kubernetes manifests.
- DGDR template — profiler and planner driven deployment request example.
Component Examples
- Router Examples — Python API usage, Kubernetes examples, and custom routing patterns.
- Planner Examples — custom load predictors and non-Kubernetes scaling environments.
- Profiler Examples — DGDR YAMLs and profiling script examples.
Repository Examples
Browse the full examples directory for source-controlled examples that may not yet have a polished docs page.