# Tokenless

> Documentation for the Tokenless API.

## Docs

- [Messages](https://docs.usetokenless.com/api-reference/anthropic/messages.md): Tokenless accepts the compatible Anthropic Messages request shape and routes the request behind a Tokenless routing profile. For the complete upstream contract, see the [Claude Messages API reference](https://platform.claude.com/docs/en/api/messages/create).
- [Models](https://docs.usetokenless.com/api-reference/anthropic/models.md): Lists Tokenless routing profiles in the compatible Anthropic model-list shape. For the upstream model API, see the [Claude Models API reference](https://platform.claude.com/docs/en/api/models/list).
- [Provider keys](https://docs.usetokenless.com/api-reference/byok/provider-keys.md)
- [Revoke provider key](https://docs.usetokenless.com/api-reference/byok/revoke-provider-key.md)
- [Set provider key](https://docs.usetokenless.com/api-reference/byok/set-provider-key.md)
- [Update provider key](https://docs.usetokenless.com/api-reference/byok/update-provider-key.md)
- [Feedback](https://docs.usetokenless.com/api-reference/feedback.md): Sets or clears the caller's rating/comment on a routed request. Send an all-empty body to clear existing feedback. The operation is idempotent.
- [Get request details](https://docs.usetokenless.com/api-reference/get-request-details.md): Fetches the routed-request record associated with `X-Request-Id` from a model response. The lookup is scoped to the authenticated user; a foreign ID returns 404.
- [List routed requests](https://docs.usetokenless.com/api-reference/list-routed-requests.md): Returns older activity-feed rows using a composite cursor. Both cursor fields are required so requests that share a timestamp are not skipped.
- [Chat Completions](https://docs.usetokenless.com/api-reference/openai/chat-completions.md): Tokenless accepts the compatible Chat Completions request shape and routes the request behind a Tokenless routing profile. For the complete upstream contract, see the [OpenAI API reference](https://developers.openai.com/api/reference/overview).
- [Models](https://docs.usetokenless.com/api-reference/openai/models.md): Lists Tokenless routing profiles in the compatible OpenAI model-list shape. For the upstream model API, see the [OpenAI API reference](https://developers.openai.com/api/reference/overview).
- [Responses](https://docs.usetokenless.com/api-reference/openai/responses.md): Tokenless accepts the compatible Responses API request shape and routes the request behind a Tokenless routing profile. For the complete upstream contract, see the [OpenAI API reference](https://developers.openai.com/api/reference/overview).
- [Session totals](https://docs.usetokenless.com/api-reference/session-totals.md)
- [Sessions](https://docs.usetokenless.com/api-reference/sessions.md)
- [Usage](https://docs.usetokenless.com/api-reference/usage.md): Returns account credit consumption, active scoped grants, and recent upstream usage events. This endpoint bypasses the quota gate so a depleted account can still inspect its balance.
- [Bring your own keys](https://docs.usetokenless.com/byok.md): Store an upstream provider key and have Tokenless use it for that provider's routed requests.
- [Anthropic-compatible API](https://docs.usetokenless.com/compatibility/anthropic.md): Use Anthropic clients or direct HTTP requests with Tokenless routing modes.
- [OpenAI-compatible API](https://docs.usetokenless.com/compatibility/openai.md): Use OpenAI clients or direct HTTP requests with Tokenless routing modes.
- [Tokenless API](https://docs.usetokenless.com/index.md): Route each request to the right frontier model through familiar OpenAI and Anthropic SDKs.
- [Quickstart](https://docs.usetokenless.com/quickstart.md): Create a key and make your first streamed request in a few minutes.
- [Errors and retries](https://docs.usetokenless.com/reference/errors.md): Handle provider-compatible errors and retry temporary failures safely.
- [Cost and usage](https://docs.usetokenless.com/reference/usage.md): See exactly what a request cost, how much credit remains, and where spend is going.
- [Routing modes](https://docs.usetokenless.com/routing-profiles.md): Select the quality, latency, and cost posture for each request.
- [Tokenless CLI](https://docs.usetokenless.com/tkl-cli.md): Install tkl, sign in through the dashboard, and run coding agents through Tokenless.

## OpenAPI Specs

- [usage](https://docs.usetokenless.com/openapi/usage.yaml)
- [compatibility](https://docs.usetokenless.com/openapi/compatibility.yaml)