> ## Documentation Index
> Fetch the complete documentation index at: https://continuum-three-olive.vercel.app/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# API overview

> Use hosted models or connect bots and computers to your application.

Continuum has two API surfaces. Use the Inference API for model requests. Use the Computer API, also called the **Continuum Bot API**, to manage bots and their computers per client.

| API                              | For                                                                                  | Credentials                                                 | Status                                          |
| -------------------------------- | ------------------------------------------------------------------------------------ | ----------------------------------------------------------- | ----------------------------------------------- |
| [Inference API](/docs/apis/inference) | Developers adding model responses, streaming, or tool calls to an app or coding tool | Inference key, `cont_sk_…`                                  | Available with eligible hosted inference access |
| [Computer API](/docs/apis/computer)   | Platforms creating bots per customer and embedding chat and computer control         | Partner key, `cont_pk_…`; browser client token, `cont_ct_…` | Preview                                         |

Inference calls keep their existing OpenAI-compatible and Anthropic-compatible endpoints. A partner key or client token does not replace an inference key.

## Get an inference key

Open **Settings → Account → Inference API** and choose **Create key**. The web app also exposes **Inference API keys** in Usage. Copy the value when it is shown; you cannot retrieve it later. Store it on your backend and revoke it from the same panel when it is no longer needed.

Connecting the Continuum provider in **Settings → Providers** creates an app-managed inference key. For your own integration, create a separate named key in Settings. See [Inference API authentication](/docs/apis/inference#authentication).

## Get Computer API access

<Note>
  The Computer API is preview. Partner provisioning, client tokens, and the embed component described here are the proposed integration contract. Confirm access and the deployed contract with your organization admin before integrating.
</Note>

The upcoming **Bots → bot action menu → Connect to your app…** sheet in the Mac app will let you choose or create a partner, allowlist its origins, issue a partner key, and connect a client and bot. It will provide an embed snippet and a quickstart for that bot. Until that flow is available, work with your **organization admin** to arrange access and key issuance. Partner keys are shown once and stay on your backend.

Your backend authenticates your application's user, checks which client they can access, and requests a short-lived client token for that client. Only that token goes to the browser. The [Computer API quickstart](/docs/apis/computer#preview-quickstart) walks through the sequence.

## Choose a starting point

<Columns cols={2}>
  <Card title="Inference API" icon="code" href="/docs/apis/inference">
    Endpoints, model discovery, effort, streaming, tool calls, and request examples.
  </Card>

  <Card title="Computer API (preview)" icon="monitor" href="/docs/apis/computer">
    Clients, bots, computer lifecycle, transcript streams, and browser embeds.
  </Card>

  <Card title="Hosted inference and billing" icon="credit-card" href="/docs/features/hosted-inference">
    Plans, included budgets, prepaid balance, and organization controls.
  </Card>

  <Card title="Providers" icon="plug" href="/docs/providers/overview">
    Connect your own provider accounts to run sessions in Continuum.
  </Card>
</Columns>