API Reference Overview

This section documents the field-by-field contract of the public endpoints: request fields, response structure, and real examples. All endpoints are authenticated via the Authorization: Bearer <API_KEY> request header, and the model name is specified with model.

Models

List Models

GET /v1/models — Lists the models available to the current tenant; list only.

Files

Upload File

POST /v1/files — Upload a file, returning a file_id for reuse in large-file scenarios.

List Files

GET /v1/files — List uploaded files.

Get File

GET /v1/files/{file_id} — Retrieve a single file's metadata by file_id.

Delete File

DELETE /v1/files/{file_id} — Delete a file by file_id.

Multimodal understanding

Multimodal Understanding

POST /v1/responses — Use MOSS-VL to understand 1–5 images or 1 video and return text.

Text to speech

Single-speaker Speech

POST /v1/audio/speech — Text-to-speech; voice accepts a voice id only.

Streaming Speech

POST /v1/audio/speech — Low-latency streaming generation with TTS 1.5 Flash; supports SSE and raw PCM.

Multi-speaker Speech

POST /v1/audio/speech/speakers — Multi-speaker TTS, using speakers to bind each speaker to a voice.

Speech recognition

Create Transcription

POST /v1/audio/transcriptions — Speech recognition, supporting file / file_id / url / audio_url inputs.

Voice management

List Voices

GET /v1/audio/voices — List the currently available voices.

Create Voice

POST /v1/audio/voices — Upload an audio sample to create a reusable voice, returning a voice_id.

Generate Voice

POST /v1/audio/voice/generations — Generate a voice / speech preview from text and an instruction.

Tasks

Get Task

GET /v1/audio/tasks/{task_id} — Query a general audio asynchronous task; successful results are expanded at the top level.