Integrations

Mostly, it is a base URL.

How do integrations work?

Serving implements the OpenAI chat-completions shape, so pointing an existing client at a fine-tuned model means changing a base URL and a key. That makes most of these pages short. What each one adds is the thing that specifically catches people out with that client — and where there is nothing, it says so.

Why there is so little to configure

A deployment serves the OpenAI chat-completions shape, so any client that already speaks it works by pointing at a different base URL with a different key. Requests, responses, streaming chunks and tool-call objects are the same structures, which means an application built against one provider moves to another without a rewrite.

That portability is deliberate and it cuts both ways: it is exactly as easy to move away from us as it is to move to us. We think a product that has to trap you is not a product worth building, and an integration layer that is a base URL is the cheapest possible proof of that.

Two things recur across these pages and are worth knowing before you start. Set temperature explicitly for anything structured — the default is not zero, and a classifier left at the default will occasionally invent. And raise your client's timeout if the endpoint can scale to zero, because a cold start is a slow first request rather than a failure, and a short timeout turns it into one.

SDKs and HTTP

OpenAI Python SDK

Python

The official Python client for the OpenAI API, and the most common way Python code talks to a language model.

OpenAI Node SDK

TypeScript

The official TypeScript and JavaScript client for the OpenAI API.

curl and plain HTTP

Shell

The lowest common denominator: an HTTP request with a JSON body, from a shell or from any language.

Frameworks

LangChain

Python

A framework for composing language-model calls into chains, agents and retrieval pipelines.

LlamaIndex

Python

A framework for connecting language models to your own data, with a focus on indexing and retrieval.

Vercel AI SDK

TypeScript

A TypeScript toolkit for building streaming language-model interfaces in web applications.

Pydantic AI

Python

A Python agent framework that uses Pydantic models to type and validate model output.

Tooling

LiteLLM

YAML

A proxy and library that presents many model providers behind one OpenAI-compatible interface.

n8n

Configuration

A workflow automation tool with a visual editor, self-hostable, with language-model nodes built in.

Open WebUI

Configuration

A self-hosted chat interface that can connect to any OpenAI-compatible backend.

Continue

JSON

An open-source coding assistant extension for VS Code and JetBrains that can use any compatible model.

Model hubs

Hugging Face Hub

HTTP

The model hub the open-weight ecosystem is organised around, for both base models and published fine-tunes.

Start with the free tier

A magic link creates your account, your tenant and your first API key. No card until you ask for compute.

Get your API key Read the docs