CosyVoice 3 Flash

Overview

CosyVoice 3 Flash is an Alibaba audio model for text-to-speech plus custom voice creation. Use this page when you are ready to call the API after trying the model in the APIXO playground.

Capability	Value
Model ID	`cosyvoice-3-flash`
Modes	`speech`, `clone`, `design`
Built-in voices	Yes
Custom voices	Yes
Speech prompt length	1-20000 characters
Design prompt length	1-500 characters
Preview text length	1-200 characters
Clone audio URLs	Exactly 1 URL

Endpoint and authentication

Base URL:

https://api.apixo.ai/api/v1

Method	Endpoint	Purpose
`POST`	`/generateTask/cosyvoice-3-flash`	Submit a speech, clone, or design task
`GET`	`/statusTask/cosyvoice-3-flash?taskId={taskId}`	Poll task status and retrieve results

All requests require your APIXO API key:

Authorization: Bearer YOUR_API_KEY

Submit requests also require:

Content-Type: application/json

Copy-paste async quickstart

This minimal request submits a speech task and returns a taskId.

curl -X POST "https://api.apixo.ai/api/v1/generateTask/cosyvoice-3-flash" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "request_type": "async",
    "input": {
      "mode": "speech",
      "voice": "longanlang_v3",
      "prompt": "Hello, welcome to APIXO.",
      "format": "mp3"
    }
  }'

Successful response:

{
  "code": 200,
  "message": "success",
  "data": {
    "taskId": "task_12345678"
  }
}

Save the taskId; you need it to poll for the final result.

Poll for result

curl -X GET "https://api.apixo.ai/api/v1/statusTask/cosyvoice-3-flash?taskId=task_12345678" \
  -H "Authorization: Bearer YOUR_API_KEY"

Processing response:

{
  "code": 200,
  "message": "success",
  "data": {
    "taskId": "task_12345678",
    "state": "processing",
    "createTime": 1767965610929
  }
}

Success response:

{
  "code": 200,
  "message": "success",
  "data": {
    "taskId": "task_12345678",
    "state": "success",
    "resultJson": "{\"resultUrls\":[\"https://file.apixo.ai/xxx.mp3\"]}",
    "createTime": 1767965610929,
    "completeTime": 1767965622450,
    "costTime": 11521
  }
}

Custom voice success response:

{
  "code": 200,
  "message": "success",
  "data": {
    "taskId": "task_12345678",
    "state": "success",
    "voice_id": "cosyvoice-v3-flash-xxxx",
    "voice_status": "OK",
    "resultJson": "{\"resultUrls\":[\"https://file.apixo.ai/preview.wav\"]}",
    "createTime": 1767965610929,
    "completeTime": 1767965682450,
    "costTime": 71521
  }
}

Failed response:

{
  "code": 200,
  "message": "success",
  "data": {
    "taskId": "task_12345678",
    "state": "failed",
    "failCode": "UNDEPLOYED",
    "failMsg": "Voice review failed. This custom voice is unavailable (UNDEPLOYED).",
    "voice_id": "cosyvoice-v3-flash-xxxx",
    "voice_status": "UNDEPLOYED",
    "createTime": 1767965610929,
    "completeTime": 1767965682450
  }
}

Parse resultJson after state becomes success:

const payload = JSON.parse(data.resultJson);
const audioUrls = payload.resultUrls;

Request body

Speech

{
  "request_type": "async",
  "input": {
    "mode": "speech",
    "voice": "longanlang_v3",
    "prompt": "Welcome to APIXO.",
    "format": "mp3",
    "sample_rate": 22050
  }
}

Clone

{
  "request_type": "async",
  "input": {
    "mode": "clone",
    "prefix": "demo01",
    "audio_urls": [
      "https://example.com/reference.wav"
    ]
  }
}

Design

{
  "request_type": "async",
  "input": {
    "mode": "design",
    "prefix": "demo02",
    "prompt": "A warm and clear female narration voice for tutorials.",
    "preview_text": "Hello, this is the preview sample.",
    "format": "wav"
  }
}

Parameters

string

default:"async"

required

Result delivery mode. Use async for polling with statusTask, or callback for webhook delivery.

string

Required when request_type is callback. Must be a public HTTPS URL that can receive the final task payload. See Webhooks.

object

required

CosyVoice 3 Flash input parameters.

Show properties

string

required

Workflow type. Supported values: speech, clone, design.

string

Required for speech. Accepts either a built-in system voice or a previously created voice_id.

string

Required for speech and design. In speech, this is the synthesis text and supports 1-20000 characters. In design, this is the voice description and supports 1-500 characters.

string[]

Required for clone. Must contain exactly 1 non-empty audio URL.

string

Required for design. Preview text for the generated custom voice. Supports 1-200 characters.

string

Required for clone and design. Must match ^[A-Za-z0-9]{1,10}$.

string

Optional output format. Passed through for speech. For design, APIXO also uses it as the preview response_format. If omitted in design, the preview defaults to wav.

integer

Optional sample rate. Passed through for speech, and also supported for clone or design preview output.

number

Optional speaking-rate control for speech.

number

Optional pitch control for speech.

number

Optional volume control for speech.

string

Optional instruction text for speech. Must not exceed 100 characters.

string[]

Optional language hints. APIXO accepts either a string or string array, then keeps only the first non-empty item before sending upstream.

boolean

Optional SSML toggle for speech.

boolean

Optional word timestamp toggle for speech.

boolean

Optional AIGC tag toggle for speech.

string

Optional AIGC propagation metadata for speech.

string

Optional AIGC propagation ID for speech.

boolean

Optional speech passthrough flag supported by the current route.

boolean

Optional Markdown filtering toggle for speech.

Response format

Submit task response

POST /generateTask/cosyvoice-3-flash returns a task ID when the task is accepted:

integer

API status code. 200 means the task was accepted.

string

Human-readable status message.

string

Unique task identifier used with the status endpoint.

Status response fields

string

Unique task identifier.

string

Current task state: pending, processing, success, or failed.

string

JSON string containing audio result URLs. Present when audio output is available.

string

Custom voice ID returned by clone or design workflows.

string

Upstream custom voice status such as DEPLOYING, OK, or UNDEPLOYED.

string

Machine-readable failure code. Present when state is failed.

string

Human-readable failure message. Present when state is failed.

integer

Task creation timestamp in Unix milliseconds.

integer

Task completion timestamp in Unix milliseconds. Present after completion.

integer

Processing duration in milliseconds. Present after successful completion.

Webhook callback mode

Use callback mode when your backend should receive the final result automatically instead of polling.

curl -X POST "https://api.apixo.ai/api/v1/generateTask/cosyvoice-3-flash" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "request_type": "callback",
    "callback_url": "https://your-server.com/webhooks/apixo",
    "input": {
      "mode": "speech",
      "voice": "longanyang",
      "prompt": "Please welcome everyone to the event."
    }
  }'

See Webhooks for delivery requirements and retry behavior.

Billing

CosyVoice 3 Flash uses different billing units by workflow.

Workflow	APIXO price
`speech`	`$0.10 / 10K characters`
`clone`	`$0.002 / request`
`design`	`$0.03 / request`

For current route and market comparison pricing, see Pricing.

Latency and polling

Actual latency may vary by text length, voice route, queue load, and whether you are creating a custom voice.

Workflow	Typical generation time	Recommended first poll	Poll interval
`speech`	5s-30s	5s after task creation	3s-5s
`clone`	30s-180s	20s after task creation	5s-10s
`design`	30s-180s	20s after task creation	5s-10s

Use callback mode for production voice creation workflows so your backend does not need to poll during longer DEPLOYING periods.

Custom voice enrollment can remain in processing while the upstream voice status is still DEPLOYING. Custom voice_id records are garbage-collected if they are not used for 7 consecutive days. If you plan to keep a custom voice active, call speech with that voice_id at least once within every 7-day window.

Errors and troubleshooting

HTTP errors

Code	Meaning	What to do
`400`	Invalid request body or parameter shape	Fix the request before retrying
`401`	Missing or invalid API key	Check the `Authorization` header
`402`	Insufficient balance or quota	Add balance or switch account/key
`403`	Key or route cannot access the model	Check permissions and route strategy
`429`	Rate limit or concurrency limit reached	Retry with exponential backoff
`500`	Server error	Retry with backoff
`502`	Upstream provider error	Retry with backoff
`504`	Upstream timeout	Retry or use callback mode

Task failure cases

Fail code	Meaning	What to do
`UNDEPLOYED`	Voice review failed and the custom voice is unavailable	Create a new voice with better input audio or a different design prompt
`TASK_TIMEOUT`	Custom voice deployment stayed in `DEPLOYING` for too long	Retry later
`VALIDATION_ERROR`	Input failed APIXO validation, or a `voice_id` is invalid for the current user/provider/model	Fix the input before retrying
`UpstreamError`	Upstream route returned an unmapped failure	Retry with backoff

Common validation rules:

voice is required for speech. APIXO also accepts legacy voice_id in the request and normalizes it into voice.
Custom voice_id values must belong to the current user, match the current provider, and match the target model.
Custom voice_id values are removed after 7 consecutive days without use.
Only cosyvoice-3-flash supports built-in system voices such as longanlang_v3, longanyang, and loongabby_v3.
audio_urls must be an array with exactly one non-empty URL.
prefix must be 1-10 letters or digits.
instruction must be at most 100 characters.

Getting Started

Image Models

Video Models

Audio Models

Text & Utility Models

CosyVoice 3 Flash

Overview

Endpoint and authentication

Copy-paste async quickstart

Poll for result

Request body

Speech

Clone

Design

Parameters

Response format

Submit task response

Status response fields

Webhook callback mode

Billing

Latency and polling

Errors and troubleshooting

HTTP errors

Task failure cases

​Overview

​Endpoint and authentication

​Copy-paste async quickstart

​Poll for result

​Request body

​Speech

​Clone

​Design

​Parameters

​Response format

​Submit task response

​Status response fields

​Webhook callback mode

​Billing

​Latency and polling

​Errors and troubleshooting

​HTTP errors

​Task failure cases

​Related links

Overview

Endpoint and authentication

Copy-paste async quickstart

Poll for result

Request body

Speech

Clone

Design

Parameters

Response format

Submit task response

Status response fields

Webhook callback mode

Billing

Latency and polling

Errors and troubleshooting

HTTP errors

Task failure cases

Related links