App service
Services/vLLM

vLLM service

vLLM inference service for serving language models through an OpenAI-compatible API on existing GPU-ready Kubernetes infrastructure.

Use it as a reusable, versioned service inside custom or predefined stacks so teams can keep the runtime and operational model consistent.

All services

Why teams use this service

Technical profile

Review versions, ports, source, and packaging.

Stacks that use it

See where this service fits in ready-made stacks.

Runtime wiring and defaults

See ports, links, storage, configs, and runtime behavior.

Overview

Technical profile and packaging

This page shows how vLLM is packaged and versioned as a Wodby service, what it exposes at runtime, and which source artifacts it builds on.

Service type

App service

Versions

0.29

Ports and protocols

8000/HTTP

Labels

vllm, inference, openai-compatible

Persistent volumes

Model cache

Stacks

Use it inside ready-made stacks or your own

These public stacks already include vLLM as part of their default service model. You can start there and customize the stack, or compose your own stack around the same building block.
vLLM

Runtime wiring

Reuse connectivity, storage, and runtime defaults

This service can expose ports, connect to other services, attach persistent storage, and define runtime defaults that become reusable inside stacks rather than being rebuilt case by case.

Endpoints

8000/HTTP

Technical domains

HTTP endpoints from this service can use technical domains with auto-renewed TLS certificates on Wodby.

Volumes

Model cache

Settings

7 settings can be overridden per environment.