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

# Build with Limitless

> Use open-weight models through the OpenAI SDK.

Limitless provides an OpenAI-compatible inference API for abliterated open-weight models. Keep the SDK you already use, configure the Limitless base URL, and authenticate with a Limitless customer key.

<Note>
  The [Limitless console](https://trylimitless.ai/app) and API are available at `https://trylimitless.ai`. Set your OpenAI SDK base URL to `https://trylimitless.ai/v1`. The current site is a sandbox beta: checkout uses test payments, while inference requests reach real models and consume your sandbox credits.
</Note>

<CardGroup cols={2}>
  <Card title="Make your first request" icon="terminal" href="/quickstart">
    Create a key and run JavaScript, Python, or cURL.
  </Card>

  <Card title="Stream a response" icon="bolt" href="/guides/streaming">
    Display text as it arrives and capture final token usage.
  </Card>

  <Card title="Understand compatibility" icon="code" href="/guides/compatibility">
    See which endpoints and request options are supported.
  </Card>

  <Card title="Track usage" icon="chart-line" href="/usage">
    Inspect requests, outcomes, token counts, and charges.
  </Card>
</CardGroup>

## What you can build

* Chat applications using `POST /v1/chat/completions`.
* Streaming interfaces using server-sent events.
* Applications that define their own tools and execute model-requested function calls.
* Stateless workflows using `POST /v1/responses`.
* Anthropic-compatible clients using `POST /v1/messages` and input-token counting.
* Applications using hosted web search, model-specific reasoning, schemas, and image inputs.
* Workspace automation for projects, scoped credentials, reports, spending caps, and versioned policies.

Model features depend on the selected model and inference backend. Consult the [compatibility guide](/guides/compatibility) before enabling tools, structured output, or multimodal inputs.

## How requests work

Your application sends its customer key to Limitless. Limitless verifies the key, checks available credits and limits, records request metadata, and forwards the request to the configured inference backend. Your application does not need an upstream provider key.

Use a server-side client. Keys grant access to your workspace's inference capacity and must not be embedded in browser or mobile application bundles.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.