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

# Tokenless API

> Route each request to the right frontier model through familiar OpenAI and Anthropic SDKs.

Tokenless is an OpenAI- and Anthropic-compatible routing API. Your application
sends a routing mode instead of selecting a provider model; Tokenless chooses
the underlying model and returns the response in the wire format you already use.

## Start in three steps

1. Create an API key in the [Tokenless dashboard](https://dashboard.usetokenless.com/access). The full key is shown once, so
   store it as a secret.
2. Point the OpenAI or Anthropic SDK at Tokenless.
3. Send `tokenless-pro` as the model for the default routing mode.

| Compatibility surface | Base URL                                 |
| --------------------- | ---------------------------------------- |
| OpenAI                | `https://api.usetokenless.com/openai/v1` |
| Anthropic             | `https://api.usetokenless.com/anthropic` |

<Card title="Quickstart" icon="rocket" href="/quickstart">
  Make a streamed request with your first Tokenless API key.
</Card>

## What stays compatible

Use the standard SDK streaming and response parsers. Tokenless supports Chat
Completions, Responses, and Anthropic Messages, including tool calls and image
inputs. Both streaming and non-streaming requests are accepted.

This documentation does not provide an interactive API playground. Examples are
copyable, but the docs site never accepts a key or sends a request on your behalf.

## Choose a routing mode

* `tokenless-pro` balances answer quality, latency, and cost for most tasks.
* `tokenless-max` favors stronger candidates for difficult, high-value work.
* `tokenless-ultra-saver` prioritizes lower cost and higher throughput.