One endpoint, one key, every model

The API speaks OpenAI's dialect, so any tool that lets you set a base URL and a key already works. No plugin to install, no fork to maintain — change two settings and the tool keeps behaving exactly as it did.

Get a key Browse 72 models

Cursor

Set an OpenAI-compatible base URL in Settings → Models, then use our model ids.

Editor

Base URL:  https://api.aiomodels.dev/v1
API key:   sk_live_...
Model:     claude-opus-5

Cline

Choose the OpenAI Compatible provider and paste the base URL and key.

VS Code extension

Provider:  OpenAI Compatible
Base URL:  https://api.aiomodels.dev/v1
API key:   sk_live_...

Continue

Add a model block to config.json with the openai provider.

VS Code / JetBrains

{
  "models": [{
    "title": "aiomodels",
    "provider": "openai",
    "apiBase": "https://api.aiomodels.dev/v1",
    "apiKey": "sk_live_...",
    "model": "claude-opus-5"
  }]
}

Aider

Two environment variables and the model flag.

Terminal

export OPENAI_API_BASE=https://api.aiomodels.dev/v1
export OPENAI_API_KEY=sk_live_...

aider --model openai/claude-opus-5

Zed

Add an OpenAI-compatible provider in settings.json.

Editor

"language_models": {
  "openai": {
    "api_url": "https://api.aiomodels.dev/v1",
    "available_models": [
      { "name": "claude-opus-5", "max_tokens": 200000 }
    ]
  }
}

Open WebUI

Add us as an OpenAI API connection; every model appears in the picker.

Self-hosted chat

Settings → Connections → OpenAI API
URL:  https://api.aiomodels.dev/v1
Key:  sk_live_...

LangChain

Use ChatOpenAI with our base URL — the rest of your chain is unchanged.

Framework

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url="https://api.aiomodels.dev/v1",
    api_key="sk_live_...",
    model="claude-opus-5",
)

Vercel AI SDK

Create an OpenAI-compatible provider and stream as usual.

TypeScript

import { createOpenAI } from "@ai-sdk/openai";

const aiomodels = createOpenAI({
  baseURL: "https://api.aiomodels.dev/v1",
  apiKey: process.env.AIOMODELS_API_KEY,
});

Anything else

If a tool accepts an OpenAI base URL, it works. Point it at https://api.aiomodels.dev/v1, use an sk_live_ key, and pick a model id from the catalog. Streaming, tool calls and usage reporting behave the way your client expects.

Read the docs