Fun-Music v1

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit

Not collected

Maximum input tokens

Not collected

Maximum output

Not collected

Input modalities

TextSource for Input modalities (opens in a new tab)

Output modalities

AudioSource for Output modalities (opens in a new tab)

Reasoning

Not collected

Tool calling

Not supportedSource for Tool calling (opens in a new tab)

Structured output

Not supportedSource for Structured output (opens in a new tab)

Streaming

SupportedSource for Streaming (opens in a new tab)

Set X-DashScope-SSE: enable. Intermediate events contain Base64 audio chunks; the final event includes the full audio URL.

Batch processing

Not supportedSource for Batch processing (opens in a new tab)

Fine-tuning

Not supportedSource for Fine-tuning (opens in a new tab)

Knowledge cutoff

Not collected

Release date

2026-05-06Source for Release date (opens in a new tab)

Documented Beijing Model Studio launch, not the current card update date.

Parameters

Not collected

Architecture

Not collected

Languages

Chinese, EnglishSource for Languages (opens in a new tab)

Lyrics support Chinese and English; other lyric languages are not supported in the reviewed contract.

Context caching

Not supportedSource for Context caching (opens in a new tab)

Search grounding

Not supportedSource for Search grounding (opens in a new tab)

Request mode

HTTPS with optional SSE audio streamingSource for Request mode (opens in a new tab)

Streaming and non-streaming modes use the same model identity, not separate checkpoints.

Maximum prompt characters

2,000Source for Maximum prompt characters (opens in a new tab)

Non-streaming prompt maximum only. Lyrics and streaming prompts have separate limits; this is not a token window.

Input requirements

Supply prompt or lyrics; when both are supplied, lyrics takes prioritySource for Input requirements (opens in a new tab)

Lyrics limits

Non-streaming: 5–350 Chinese characters or 5–2,000 English characters. Streaming: 300–350 Chinese characters or 200–250 English words.Source for Lyrics limits (opens in a new tab)

Streaming text limits

Prompt: 5–1,000 Chinese characters or English wordsSource for Streaming text limits (opens in a new tab)

The general guide says characters for English too, whereas the API parameter table says English words. This discrepancy remains unresolved; do not treat the larger word allowance as a universal guarantee.

Music controls

Prompt describes style, scene, mood and instruments; male/female vocal selector, default femaleSource for Music controls (opens in a new tab)

Described preferences guide generation; there is no verified deterministic tempo or instrument guarantee.

Vocal selection

SupportedSource for Vocal selection (opens in a new tab)

The gender parameter is supported only by fun-music-v1. Both models can produce vocals; this flag concerns explicit selection.

Instrumental generation

SupportedSource for Instrumental generation (opens in a new tab)

The domestic API documents is_instrumental=true, which ignores lyrics and gender. The international API page omits this field; account-specific support should be confirmed.

Output formats

MP3, WAVSource for Output formats (opens in a new tab)

MP3 is the default. Sample rate and channel count are returned as metadata, not documented configurable guarantees.

Audio identification marker

Optional audible Morse-code AI marker appended to the generated audioSource for Audio identification marker (opens in a new tab)

enable_aigc_watermark defaults to false. Enabling it increases output duration; do not describe this as an invisible watermark.

Result and task expiry

Complete audio download URL valid for 24 hoursSource for Result and task expiry (opens in a new tab)

URL expiry does not establish deletion of all retained request data.

Music duration

Not collected

No universal maximum duration is documented in the reviewed contract. The 200-second response example is not a limit.

Audio sample rates

Not collected

The API shows 48,000 Hz as example response metadata, not a guaranteed fixed or selectable output rate.

Audio channels

Not collected

Stereo appears in the example response, but the reviewed API does not establish a universal channel-count guarantee.

Usage metering

Generated audio seconds; input is freeSource for Usage metering (opens in a new tab)

The API usage.duration field reports billable music duration. Free input is not free generation. No generic text-question tariff is inferred.

Data retention

Standard API inference prompts and responses are retained up to 30 days, subject to legal/security exceptions; stateful files remain until deleted. Fun-Music is absent from the published ZDR-supported model list. A 24-hour download-link lifetime is not a data-deletion guarantee.Source for Data retention (opens in a new tab)

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

Why it is included

Inference from the distinct approval-gated hosted selectors, their official API/model cards and review of the public FunMusic/InspireMusic repository and Qwen/FunAudioLLM inventories. No matching hosted-selector weight release or checkpoint mapping was verified. The public InspireMusic checkpoints remain separate; neither this classification nor the hosted name makes the whole research family closed-source.

Lifecycle dates

Sources reviewed