Docs

Google FlowMusic Lyria 3.5 API developer guide

Google FlowMusic Lyria 3.5 is currently "Available". This guide provides model IDs, price snapshots, and copyable examples for integration. Google FlowMusic Lyria 3.5 model page

Overview

Vendor: Google
Type: Audio
Number of tiers: 1
Endpoint tags: audio-task

Available tiers and pricing

The model in requests must use the model IDs in the table below, case-sensitive.

Tier	Model ID	Input	Output	Context
Google FlowMusic Lyria 3.5	`flowmusic-lyria-3.5`	$0.033000 / 次	总价按请求参数和实际用量计算	—

Prices are build-time snapshots; refer to the pricing page and console for the latest prices.

Base URL and authentication

Base URL is https://api.rokoapi.com/v1. All requests must include Authorization: Bearer $OPENAI_API_KEY.

Console

Quickstart

curl https://api.rokoapi.com/v1/audios/tasks \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "flowmusic-lyria-3.5",
  "prompt": "cinematic ambient music with warm strings and a gentle piano",
  "duration": 30,
  "metadata": {
    "lyrics": "[Verse]\nNeon rain falls softly tonight",
    "title": "Neon Rain",
    "bpm": 120,
    "seed": "42"
  }
}'

Common parameters

Parameter	Description
`model`	Use the public model IDs from the table above.
`prompt` / `messages`	Input prompt, conversation messages, or generation instructions; format depends on the endpoint.
`prompt`	Generation models can accept size, quality, duration, aspect ratio, or reference asset URLs based on model capabilities.
`metadata.lyrics`	Generation models can accept size, quality, duration, aspect ratio, or reference asset URLs based on model capabilities.
`metadata.title`	Generation models can accept size, quality, duration, aspect ratio, or reference asset URLs based on model capabilities.
`duration`	Generation models can accept size, quality, duration, aspect ratio, or reference asset URLs based on model capabilities.
`metadata.bpm`	Generation models can accept size, quality, duration, aspect ratio, or reference asset URLs based on model capabilities.
`metadata.seed`	Generation models can accept size, quality, duration, aspect ratio, or reference asset URLs based on model capabilities.

After a successful async generation task submission, a task ID is returned. Poll GET /v1/tasks/{task_id} until the status becomes completed or failed.

Notes

Use only the model IDs listed on this page; unprovisioned or retired names will return a model unavailable error.
If you encounter a 403 error, check your account balance, key group, and model access permissions.
Multimodal, tool calling, caching, and streaming capabilities depend on the upstream model and channel configuration; refer to actual responses.