# TOKEN

> TOKEN serves frontier and open AI models through one OpenAI-compatible API at https://api.thetoken.dev/v1, with automatic failover between providers and prepaid, pay-as-you-go credit. It also serves the Anthropic Messages API at https://api.thetoken.dev/v1/messages.

Authenticate with `Authorization: Bearer $TOKEN_API_KEY`, a key created in the console (https://app.thetoken.dev). Use the model IDs listed in https://thetoken.dev/models.json. Every docs page below is also available as one file: https://thetoken.dev/llms-full.txt.

## Get started

- [TOKEN documentation](https://thetoken.dev/docs.md): Call TOKEN's models through one OpenAI-compatible API. Start here.
- [Quickstart](https://thetoken.dev/docs/quickstart.md): Create an account and a key, then make your first TOKEN request in curl, Python or TypeScript.

## Guides

- [Guides](https://thetoken.dev/docs/coding-tools.md): Use TOKEN models in coding agents, editors and SDKs.
- [Claude Code](https://thetoken.dev/docs/guides/claude-code.md): Run Claude Code on TOKEN models through the Anthropic Messages API.
- [Cursor](https://thetoken.dev/docs/guides/cursor.md): Use TOKEN models in Cursor's chat with the OpenAI API key and base URL override.
- [Cline](https://thetoken.dev/docs/guides/cline.md): Use TOKEN models in Cline through its OpenAI Compatible provider.
- [OpenCode](https://thetoken.dev/docs/guides/opencode.md): Add TOKEN as a provider in OpenCode's opencode.json.
- [OpenAI SDK](https://thetoken.dev/docs/guides/openai-sdk.md): Switch OpenAI's Python and TypeScript SDKs, or any OpenAI-compatible client, to TOKEN by changing the base URL and key.
- [LangChain](https://thetoken.dev/docs/guides/langchain.md): Use TOKEN models with LangChain's ChatOpenAI and OpenAIEmbeddings in Python and JavaScript.

## Features

- [Streaming](https://thetoken.dev/docs/features/streaming.md): Receive the reply as it is generated, as server-sent events, with token usage in the last chunk.
- [Tool calling](https://thetoken.dev/docs/features/tools.md): Let models call your functions, in the OpenAI or the Anthropic tool format.
- [Structured output](https://thetoken.dev/docs/features/structured-output.md): Get replies as JSON, or as JSON that follows your JSON Schema, with response_format.
- [Vision](https://thetoken.dev/docs/features/vision.md): Send images to models with image input, as a URL or as base64 data.
- [Embeddings](https://thetoken.dev/docs/features/embeddings.md): Turn text into vectors for search and clustering with BGE-M3 on POST /v1/embeddings.

## Reference

- [Errors and rate limits](https://thetoken.dev/docs/errors.md): Every status code TOKEN returns, the exact messages, the per-account limits, the balance reservation and how to retry.
- [API reference](https://thetoken.dev/docs/api.md): TOKEN's endpoints, request fields, responses and models, generated from the OpenAPI document and the model catalog.

## Product

- [How routing and failover work](https://thetoken.dev/docs/routing.md): Provider order, full-precision guarantee, retries and single-provider models.
- [The us-eu-hosted privacy group](https://thetoken.dev/docs/privacy-group.md): Keep requests on US and EU hosts, never on the model makers' own APIs.

## Optional

- [OpenAPI document](https://thetoken.dev/openapi.json): OpenAPI 3.1 description of the API, with the current model list
- [Model catalog](https://thetoken.dev/models.json): every model with aliases, capabilities, prices and providers, as JSON
- [Pricing](https://thetoken.dev/pricing/): prices per model, credit purchase fee, refunds
- [Status](https://status.thetoken.dev): live availability