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

# Unified API

The SLNG Unified API is a normalization layer across every speech model provider.

## The problem it solves

Every speech provider has its own way of doing things: own request format, own error handling, own parameters. When you want to move an integration from model A to model B, you usually have to build and maintain a second integration. That means one SDK per provider, different request formats, provider-specific error handling, and weeks spent evaluating each new model.

## How it works

You call a model available on the SLNG gateway, and the Unified API translates the request into whatever each provider expects, then translates the answer back.

* Authentication, request format, and responses stay the same across every model.
* To switch providers, you only change which model you point to.
* It covers both speech to text (STT) and text to speech (TTS).
* SLNG-hosted models and third-party models sit behind the same unification.

When you need a model's provider-native parameters, call its direct endpoint instead. The [Unified API guide](/guides/models/unified-api) compares the two and lists the fields each transport accepts.

## Powers our third-party integrations

The Unified API also powers our integrations with the major voice agent frameworks. In LiveKit and Pipecat, developers name the model they want, and the [SLNG plugins](/integrations/plugins/livekit) connect through the Unified API automatically. Switching providers is a change to the model name, with no other code changes.

To make your first request, follow the [Unified API guide](/guides/models/unified-api).


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