> ## Documentation Index
> Fetch the complete documentation index at: https://docs.platform.aiplanet.com/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Gateway

> One OpenAI-compatible API that routes applications across model providers with access control, limits, budgets, and usage attribution.

The AI Planet AI Gateway gives your applications a single OpenAI-compatible endpoint. Behind it, the gateway routes each request to the right provider and enforces who can call what, how much, and at what cost.

<Frame caption="The gateway console overview: request, token, and spend totals for the selected period, daily spend, request volume, top virtual keys, and requests by model.">
  <img src="https://mintcdn.com/aiplanet/HCpBly2Dz7WdHHWf/images/llm-gateway/console-overview.png?fit=max&auto=format&n=HCpBly2Dz7WdHHWf&q=85&s=7f824803f1b895dac179a326beb99f4b" alt="AI Gateway console overview page" width="1920" height="1080" data-path="images/llm-gateway/console-overview.png" />
</Frame>

## What the gateway provides

* Stable model aliases such as `fast`, `smart`, `vision`, and `embed`.
* OpenAI-compatible chat completions, Responses, embeddings, and streaming.
* Scoped API keys per organization, application, and environment.
* RPM, TPM, request-size, model, endpoint, and capability limits.
* Monetary budget enforcement with usage and cost reporting.
* Weighted routing, retries, provider fallbacks, and cooldowns.
* Prometheus metrics, readiness probes, structured logs, and budget alerts.

## Choose your path

<CardGroup cols={3}>
  <Card title="Connect an application" icon="plug" href="/llm-gateway/get-started/client-quickstart">
    Point an OpenAI SDK at an existing gateway and make your first request.
  </Card>

  <Card title="Deploy the gateway" icon="server" href="/llm-gateway/get-started/deploy">
    Run the supported Docker Compose topology in your own cloud account.
  </Card>

  <Card title="Keys and budgets" icon="key" href="/llm-gateway/administration/api-keys">
    Create scoped keys and spend budgets, by console or by API.
  </Card>
</CardGroup>

## Administration

The current release supports a private operator console plus authenticated APIs and command-line helpers. See the [Console guide](/llm-gateway/console) for model, key, budget, and usage workflows. It is not a public self-service control plane.

<CardGroup cols={2}>
  <Card title="Client API" icon="code" href="/llm-gateway/api/client-api">
    The OpenAI-compatible surface available to application keys.
  </Card>

  <Card title="Capabilities and limitations" icon="list-check" href="/llm-gateway/capabilities">
    What is supported today and what is not.
  </Card>
</CardGroup>