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

# CarHub developer documentation

> Build automotive AI workflows with the CarHub API.

CarHub exposes vehicle understanding, condition assessment, pricing and image-generation models through one API. Submit an asset or JSON request, then retrieve the job result by polling or webhook.

Every inference endpoint is independently priced and rate-limited. You can adopt a single model without committing to the rest of the platform.

<CardGroup cols={2}>
  <Card title="Quickstart" icon="rocket" href="/quickstart">Send a first request and retrieve its job.</Card>
  <Card title="API reference" icon="code" href="/api-reference/overview">Browse generated endpoints, schemas and examples.</Card>
  <Card title="Test mode" icon="flask" href="/guides/test-mode">Exercise deterministic success and failure scenarios.</Card>
  <Card title="Go live" icon="check" href="/guides/go-live">Complete the production integration checklist.</Card>
  <Card title="Player SDK" icon="cube" href="/sdk/player">Embed the interactive 360° vehicle player in your application.</Card>
</CardGroup>

## Recommended path

1. [Authenticate](/authentication) your backend with an API key.
2. Complete the [quickstart](/quickstart) with a low-risk endpoint.
3. Use [test mode](/guides/test-mode) to verify retries, empty results and failures.
4. Find the model you need in the generated [API reference](/api-reference/overview).
5. Configure [webhooks](/guides/webhooks) or poll [jobs](/guides/jobs), then use the [go-live checklist](/guides/go-live).

## Choose an integration surface

| Surface               | Use it for                                   | Authentication       |
| --------------------- | -------------------------------------------- | -------------------- |
| Inference API (`/v1`) | Models, jobs, uploads and endpoint catalogue | Organisation API key |
| MCP (`/mcp`)          | Calling active models from an AI agent       | Organisation API key |

<Note>The inference API is asynchronous by default. A successful submission usually returns `202` and a job rather than an immediate model result.</Note>