> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.fabricgateway.ai/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.fabricgateway.ai/_mcp/server.

# Discover what's available

> Find out which models, MCP servers, and APIs your gateway exposes, which path reaches each one, and whether it needs a gateway key.

Before you can send a request, you need to know what the gateway exposes and where. Each LLM provider, MCP server, and API sits behind a route with its own path, such as `/v1/billing` or `/v1/newrelic/mcp`, and a route can be open or require a gateway key. This topic shows you how to find the routes you can use, how to list the models on an LLM route, and how to tell whether a route needs a key.

## Prerequisites

Before you begin, make sure you have the following:

* The data plane host for your organization, or the base URL of at least one route. If you don't have either, ask your admin.
* A gateway key, if the routes you want to use require one. For details, see [Authenticate and send a request](/use/authenticate).

## Find the routes

If you have console access, select **LLMs**, **MCP servers**, or **Services** in the sidebar to see what your organization has connected, and then open the **Routes** tab to see each route's path. To copy a base URL, see [Get the base URL](/use/authenticate#get-the-base-url).

If you don't have console access, ask your admin for the list of route base URLs and which ones require a key.

## List the models on an LLM route

To list the models a route serves, send a request to `/v1/models`:

```bash
curl "BASE_URL/v1/models" \
  -H "fg-api-key: GATEWAY_KEY"
```

On a route with a model router, the gateway answers this itself, and the list holds every `<provider>/<model>` name and alias that your key can use. For the names and the errors, see [Choose a model](/use/llm/choose-a-model).

## Find out whether a route needs a key

Send a request without a key and read the status and the body. With no key, the gateway serves the request from the Default namespace, so this test only tells you about routes in Default. A short `Unauthorized` or `404 page not found` comes from the gateway. Any other response doesn't tell you whether the route needs a key, because the gateway also returns provider-shaped errors of its own, such as `unknown LLM endpoint` and `model_not_found`. For those, see [Troubleshoot access](/use/troubleshoot-access):

* `401` means that a route in Default requires authentication. Send a gateway key.
* `404` means that Default has no route at that path. The route might exist in another namespace. Send your gateway key, because the key, not the URL, picks the namespace. For details, see [Understand how your key picks a namespace](/use/authenticate#understand-how-your-key-picks-a-namespace).
* Any other status means the route accepted the request without a key, or it reached a later step. For what each status means, see [Troubleshoot access](/use/troubleshoot-access).

## Next steps

* [Authenticate and send a request](/use/authenticate)
* [Choose a model](/use/llm/choose-a-model)
* [Connect an MCP client](/use/mcp/connect-a-client)
* [Call a REST route](/use/call-a-rest-api)