Skip to main content
Each Moss SDK has an inference client for the Responses endpoint, separate from MossClient. Any OpenAI client also works with the base URL changed: see OpenAI compatibility.

Packages

Guides

Python

An async client in the moss package.

JavaScript

A drop-in for OpenAI’s client, plus indexes.

Browser

Call from a page, with the project key kept on your server.

Swift

responses.create, and createStreaming for events.

Android

A Kotlin client to call from Dispatchers.IO.

Elixir

Moss.Inference.Responses.create/3, from any process.

C

Requests and responses as JSON, with an event callback.