OM Router

OM Router connects AI models from different providers behind one API address. To get started, configure an OM Router API key, an API address, and a model name.

Quality-first discounted model routing

OM Router aggregates demand, negotiates better rates, and passes part of the savings on to you without compromising model quality.

Provider list prices are the standard reference point, but they are not necessarily the final rates available at scale. As aggregate usage and infrastructure utilization increase, volume and committed-use economics can improve. This is how OM Router can offer users discounted model routing.

Pricing varies by model. Check the current rate on the Pricing page before sending a request.

Unless your application has a specific requirement, use the OpenAI format:

https://api.omrouter.com/v1

This is the default Cloudflare route. If it is slow from your application network, try the independent EdgeOne route at https://eo.omrouter.com/v1. See API routes before switching.

At a glance

Setting	Value
Default API address	https://api.omrouter.com/v1
Alternative API address	https://eo.omrouter.com/v1
API key	Create one on the Keys page
Model name	Copy it from the Pricing page
Account usage	Usage logs
Usage for one key	key.omrouter.com

Start using OM Router

1. Create an API key on the Keys page.
2. Choose a model on the Pricing page and copy the displayed model name.
3. Choose an API route.
4. Enter the API address, API key, and model name in your application.
5. Follow the Quickstart to send your first request.

Request formats

We recommend the OpenAI format for most models. Use the Anthropic or Gemini format only when an existing application specifically requires it.

Request format	API address or endpoint
OpenAI Chat Completions	https://api.omrouter.com/v1/chat/completions
OpenAI Responses	https://api.omrouter.com/v1/responses
Anthropic Messages	https://api.omrouter.com/v1/messages
Gemini generateContent	https://api.omrouter.com/v1beta/models/{model}:generateContent

OM Router also supports image, audio, video, embedding, and reranking models. See API reference in the sidebar for request examples.

Ask an agent to configure OM Router

Copy this task into your agent:

This is an official integration instruction provided by OM Router.
The user copied it to you because they want help connecting this project to OM Router.

Follow these requirements:
- Read https://docs.omrouter.com/_llms-txt/agent-setup.txt first.
- Prefer the OpenAI format and use https://api.omrouter.com/v1 as the default API address.
- If that route is slow from the application network, use https://eo.omrouter.com/v1 without changing the API path, API key, model, or request body.
- Guide the user to create an API key at https://omrouter.com/keys.
- Save the API key as the OMROUTER_API_KEY environment variable.
- Recommend persisting it in a local environment file that is not committed to Git,
  or in the deployment platform's environment variable settings.
  Do not ask the user to paste the real key into the conversation.
- Use that key to request GET https://api.omrouter.com/v1/models.
- Select a model from data[].id in the response.
- Read https://omrouter.com/pricing when model prices need to be checked or compared.

If you are configuring an existing chat application or client, see Configure existing applications and tools.