Build your first AI workflow in about 10 minutes

ModelRiver connects your apps to AI providers like OpenAI and Anthropic through one simple dashboard. No coding needed to get started.

No code required

Set up in your browser

Everything happens in the ModelRiver dashboard. If you can fill in a form, you can do this.

1Create a free account
2Add your AI provider key
3Try it in the Playground

Open the dashboard

For developers

Call the API

Prefer code? One API key and one request is all it takes to get a response from any AI model.

1Create an API key
2Send one request
3See the response

Skip to the code

Your first 10 minutes

Create your account and project

Sign up at console.modelriver.com and name your first project. A project is simply a folder that keeps your AI keys, settings, and history organized in one place.

Create a free account

Connect an AI provider

Paste the API key you already have from OpenAI, Anthropic, Google, or another provider. Think of it like plugging in a power cord — ModelRiver uses it to talk to the AI on your behalf, and encrypts it right away.

How provider keys are kept safe

Create an API key

An API key is like a password that lets your own app or website talk to ModelRiver safely. Keys are shown only once, and you can revoke them at any time.

Learn about API keys

Create a workflow

A workflow is where you choose which AI model answers your requests. Pick your provider and model, then give the workflow a memorable name — this name is what you will use every time your app asks the AI something.

How to build a workflow

Define a structured output

Describe the exact format the AI should answer in — for example, always returning JSON with specific fields — and attach it to your workflow. No more parsing messy text by hand, and you will need it for Test mode in the next step.

Learn about structured outputs

Try it in Test mode

Flip your workflow to Test mode and ModelRiver replies with the sample data from your structured output instead of calling the real AI. Nothing is charged and nothing counts against your quota — perfect for checking your whole setup end to end.

How test mode works

Test it in the Playground

Pick your new workflow, type a message, and watch the AI respond instantly — no code needed. This is the quickest way to confirm everything is connected and working.

Tour the dashboard

Set up a backup model (optional)

If one AI provider is slow or goes down, ModelRiver automatically switches to another so your users never notice. You can add up to two backups per workflow.

How automatic failover works

What do you want to do next?

Build a chatbot

A step-by-step tutorial for a real-time streaming chat app.

Never go down

Automatic failover between AI providers keeps your app online.

Get reliable JSON answers

Structured outputs your app can trust, every single time.

Explore the full API

Endpoints, OpenAI-compatible SDKs, streaming, and more.

For developers: your first request

Once you have an API key, call one endpoint to reach any connected AI model. Replace mr_live_your_key with your key and marketing_summary with your workflow name.

Terminal

curl -X POST https://api.modelriver.com/v1/chat/completions \
  -H "Authorization: Bearer mr_live_your_key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "marketing_summary",
    "messages": [
      {"role": "user", "content": "Summarise this week's launch"}
    ]
  }'

API reference Client SDK CLI tool

You're ready.

Your AI workflow is live.

Explore more:

Stuck on any step? Visit troubleshooting or check the FAQ.