Moss API Platform

The Moss API Platform is MOSI's developer console for accessing its generative models through an API. The homepage presents the moss-image text-to-image model and the moss-tts-1.5 speech model, with a Playground for trying them and API keys for integrating them into applications.

What It Is

A web console and API service for building with MOSI models. Developers can sign in, create API keys, test models in a Playground, read API documentation, and monitor credit balance, usage, and billing standards. The homepage also describes a documentation assistant that answers questions about getting started, choosing models or endpoints, interpreting errors, and billing, available after login.

Models and Capabilities

The homepage describes moss-image as a text-to-image model supporting long prompts of up to 4K tokens, multilingual text rendering, and JSON-based layout control. The platform says its moss-tts-1.5 model offers voice cloning, continuation generation, pause control, and pronunciation correction. Documented scenarios include audio transcription, single-voice TTS, and voice design, along with a voice library on Mossland.

Quickstart Workflow

The quickstart shows a POST request to the /v1/audio/speech endpoint with a Bearer API key. The request body specifies a model (moss-tts-1.5-flash), input text that can include inline pause markers such as [pause 0.8s], a voice_id, an mp3 response format, and URL-based delivery.

Current Status

The homepage banner announces that the MOSS Image API is now live and links to a Discord community. Logged-in users can claim a daily login credit grant for trying the models.