Use aiplug
HTTP server

HTTP server

aiplug serve starts a tiny HTTP server that speaks the OpenAI Chat Completions wire format. Point any OpenAI client SDK at it and switch the underlying provider with a single CLI command.

Run it

npx aiplug serve --port=3711 --host=127.0.0.1

Binds to loopback by default — the server is a local control plane, not a public service. Pass --host=0.0.0.0 to make it reachable on the LAN (and understand the implications for auth).

Endpoints

Method	Path	Notes
`POST`	`/v1/chat/completions`	OpenAI Chat Completions; SSE when `stream: true` is set in the body
`POST`	`/v1/responses`	Alias for `/v1/chat/completions`
`POST`	`/v1/embeddings`	OpenAI embeddings
`POST`	`/v1/images/generations`	OpenAI images
`POST`	`/v1/audio/speech`	OpenAI TTS
`POST`	`/v1/audio/transcriptions`	Multipart upload (Whisper-style)
`GET`	`/v1/models`	List models from the active provider
`GET`	`/healthz`	Health check

Use any OpenAI SDK against it

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'http://localhost:3711/v1',
  apiKey: 'whatever',  // aiplug handles the real auth
});

const reply = await client.chat.completions.create({
  model: 'claude-3-5-sonnet-latest',
  messages: [{ role: 'user', content: 'Hello' }],
});

The application code is identical regardless of which provider is active behind the server. Switch providers with aiplug transport use <slug> and the next request routes to a different backend — no SDK changes, no restarts needed.

Works with every OpenAI client SDK on the planet: Python openai, JS openai, Go openai-go, Java, Rust, and any tool that speaks the wire format.

When to use it

You have an existing tool that talks OpenAI’s wire format. Drop aiplug’s HTTP server in front and you get a swappable provider for free.
You want a single endpoint for local model + cloud model routing without writing glue.
You’re prototyping and want to test the same prompt across providers without changing your code.

When not to use it

You’re already in a Node.js / Bun process and can import aiplug directly. The HTTP server adds a network hop and a JSON round-trip per call.
You need provider-specific features that the OpenAI wire format doesn’t carry. Use the programmatic AIPlug API for those.

Ephemeral ports

Pass --port=0 for an OS-assigned port. Useful in tests and one-off scripts.

npx aiplug serve --port=0

The CLI prints the bound port on startup.