Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Stability AI / Stable Audio

Stable Audio 3.0 Large

The API-hosted Large member of Stable Audio 3, with variable-length music and sound generation and audio editing.

Music generationSound effectsAudio processing
availableClosed weightsMusic generationSound effects generationAudio processing
Official documentation (opens in a new tab)Reviewed 13 Oct 2026
12 of 23 specification fields collected
Audio output duration1–380 secondsSee specification notes below
Audio formatsMP3, WAVSee specification notes below
Input billingSee ratesStability AI API
Output billingSee ratesPublished units and conditions below

Specifications

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 output
Not collected
Input modalities
Text, AudioSource 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 collected
Structured output
Not collected
Streaming
Not collected
Batch processing
Not collected
Fine-tuning
Not collected
Knowledge cutoff
Not collected
Release date
2026-05-20Source for Release date (opens in a new tab)
Parameters
2.7 billion DiT parametersSource for Parameters (opens in a new tab)

Backbone only. Autoencoder and other components are separate.

Architecture
Latent diffusion transformer using SAME semantic-acoustic audio latents, flow-matching pretraining and adversarial post-trainingSource for Architecture (opens in a new tab)
Languages
Not collected
Context caching
Not collected
Sample rate
44.1 kHz stereoSource for Sample rate (opens in a new tab)
Input file formats
MP3, WAVSource for Input file formats (opens in a new tab)
Output duration
1–380 secondsSource for Output duration (opens in a new tab)

Request schema limit. Rounded marketing descriptions say six or three minutes respectively.

Maximum prompt characters
10,000 charactersSource for Maximum prompt characters (opens in a new tab)
Output formats
MP3, WAVSource for Output formats (opens in a new tab)
Audio duration limits
6–380 seconds for input audioSource for Audio duration limits (opens in a new tab)
Generation modes
text-to-audio, audio-to-audio, inpaintSource for Generation modes (opens in a new tab)

Access & pricing

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

Stability AI API

Provider source (opens in a new tab)
API model ID
stable-audio-3
Access
Developer account, API key and credits required. Public documentation and contracts reviewed; no inference request was made.
Regions
Not established for this endpoint; account availability is not a data-residency guarantee.
API endpoint URL
POST https://api.stability.ai/v2beta/audio/stable-audio/text-to-audioSource for API endpoint URL (opens in a new tab)
Request mode
Asynchronous generation with result pollingSource for Request mode (opens in a new tab)

JSON or base64 transport does not establish structured model output, streaming or a batch API.

Required request fields
promptSource for Required request fields (opens in a new tab)
Maximum prompt characters
10,000 charactersSource for Maximum prompt characters (opens in a new tab)
Output formats
MP3, WAVSource for Output formats (opens in a new tab)
Seed control
0–4,294,967,294; zero or an omitted seed selects randomnessSource for Seed control (opens in a new tab)
Output duration
1–380 secondsSource for Output duration (opens in a new tab)

Request schema limit. Rounded marketing descriptions say six or three minutes respectively.

Input modalities
TextSource for Input modalities (opens in a new tab)
Output modalities
AudioSource for Output modalities (opens in a new tab)
Sample rate
44.1 kHz stereoSource for Sample rate (opens in a new tab)
Sampling steps
4–8; default 8Source for Sampling steps (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Audiotext-to-audio1$0.08successful generation

Rate conditions

  1. 8 credits × published $0.01/credit. Flat rate for the selected model per successful result; duration does not change this published quote. No text-token or per-question conversion.
Pricing source (opens in a new tab)

Data & enterprise terms

Stability terms permit using content to improve services and describe a training opt-out. This is not a default no-training or zero-retention guarantee. Partner contracts may differ; enterprise terms must be verified separately.

Read the service policy (opens in a new tab)

Stability AI API

Provider source (opens in a new tab)
API model ID
stable-audio-3
Access
Developer account, API key and credits required. Public documentation and contracts reviewed; no inference request was made.
Regions
Not established for this endpoint; account availability is not a data-residency guarantee.
API endpoint URL
POST https://api.stability.ai/v2beta/audio/stable-audio/audio-to-audioSource for API endpoint URL (opens in a new tab)
Request mode
Asynchronous generation with result pollingSource for Request mode (opens in a new tab)

JSON or base64 transport does not establish structured model output, streaming or a batch API.

Required request fields
prompt, audioSource for Required request fields (opens in a new tab)
Maximum prompt characters
10,000 charactersSource for Maximum prompt characters (opens in a new tab)
Output formats
MP3, WAVSource for Output formats (opens in a new tab)
Seed control
0–4,294,967,294; zero or an omitted seed selects randomnessSource for Seed control (opens in a new tab)
Output duration
1–380 secondsSource for Output duration (opens in a new tab)

Request schema limit. Rounded marketing descriptions say six or three minutes respectively.

Input modalities
Text, AudioSource for Input modalities (opens in a new tab)
Output modalities
AudioSource for Output modalities (opens in a new tab)
Sample rate
44.1 kHz stereoSource for Sample rate (opens in a new tab)
Audio duration limits
6–380 seconds for input audioSource for Audio duration limits (opens in a new tab)
Input file formats
MP3, WAVSource for Input file formats (opens in a new tab)
Request limits
100 MB maximum request sizeSource for Request limits (opens in a new tab)

The documentation writes the suffix as Mb; no different binary megabyte convention is inferred.

Sampling steps
4–8; default 8Source for Sampling steps (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Audioaudio-to-audio1$0.08successful generation

Rate conditions

  1. 8 credits × published $0.01/credit. Flat rate for the selected model per successful result; duration does not change this published quote. No text-token or per-question conversion.
Pricing source (opens in a new tab)

Data & enterprise terms

Stability terms permit using content to improve services and describe a training opt-out. This is not a default no-training or zero-retention guarantee. Partner contracts may differ; enterprise terms must be verified separately.

Read the service policy (opens in a new tab)

Stability AI API

Provider source (opens in a new tab)
API model ID
stable-audio-3
Access
Developer account, API key and credits required. Public documentation and contracts reviewed; no inference request was made.
Regions
Not established for this endpoint; account availability is not a data-residency guarantee.
API endpoint URL
POST https://api.stability.ai/v2beta/audio/stable-audio/inpaintSource for API endpoint URL (opens in a new tab)
Request mode
Asynchronous generation with result pollingSource for Request mode (opens in a new tab)

JSON or base64 transport does not establish structured model output, streaming or a batch API.

Required request fields
prompt, audioSource for Required request fields (opens in a new tab)
Maximum prompt characters
10,000 charactersSource for Maximum prompt characters (opens in a new tab)
Output formats
MP3, WAVSource for Output formats (opens in a new tab)
Seed control
0–4,294,967,294; zero or an omitted seed selects randomnessSource for Seed control (opens in a new tab)
Output duration
1–380 secondsSource for Output duration (opens in a new tab)

Request schema limit. Rounded marketing descriptions say six or three minutes respectively.

Input modalities
Text, AudioSource for Input modalities (opens in a new tab)
Output modalities
AudioSource for Output modalities (opens in a new tab)
Sample rate
44.1 kHz stereoSource for Sample rate (opens in a new tab)
Audio duration limits
6–380 seconds for input audioSource for Audio duration limits (opens in a new tab)
Input file formats
MP3, WAVSource for Input file formats (opens in a new tab)
Request limits
100 MB maximum request sizeSource for Request limits (opens in a new tab)

The documentation writes the suffix as Mb; no different binary megabyte convention is inferred.

Sampling steps
4–8; default 8Source for Sampling steps (opens in a new tab)
Audio infill
One mask_start/mask_end interval within 0–380 secondsSource for Audio infill (opens in a new tab)

The research discusses more editing modes, but the current API exposes one interval.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Audioinpaint1$0.08successful generation

Rate conditions

  1. 8 credits × published $0.01/credit. Flat rate for the selected model per successful result; duration does not change this published quote. No text-token or per-question conversion.
Pricing source (opens in a new tab)

Data & enterprise terms

Stability terms permit using content to improve services and describe a training opt-out. This is not a default no-training or zero-retention guarantee. Partner contracts may differ; enterprise terms must be verified separately.

Read the service policy (opens in a new tab)

Stability enterprise deployment

Provider source (opens in a new tab)
API model ID
Not collected

No pinned model selector established for this access path.

Access
Large is offered for negotiated enterprise self-hosting. Public Small/Medium LoRA code does not establish local Large fine-tuning support.
Regions
Not established for this endpoint; account availability is not a data-residency guarantee.

Endpoint pricing has not been collected. Check the provider for applicable rates.

Data & enterprise terms

Stability terms permit using content to improve services and describe a training opt-out. This is not a default no-training or zero-retention guarantee. Partner contracts may differ; enterprise terms must be verified separately.

Read the service policy (opens in a new tab)

Caching

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

01 / INITIAL INPUTProcess & retain

The system processes eligible input and retains reusable state.

02 / LATER REQUESTCheck available state

Reuse depends on matching input, session state and the cache policy.

03 / CACHE HITReuse & generate

Reusing state can reduce repeated work. Any charges follow the billing rules below.

Stability AI API

stable-audio-3
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Stability AI API

stable-audio-3
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Stability AI API

stable-audio-3
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Stability enterprise deployment

Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

No verified cache rate for this endpoint. An empty rate is not a zero charge.

What the measurements say

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.

The record behind the model

Why it is included

Stability explicitly places Large outside the Stable Audio 3 open-weight release. The technical report marks Large weights unavailable, while Small and Medium weights are released. Enterprise self-hosting is negotiated access, not a public weight release.

Classification evidence (opens in a new tab)

Lifecycle dates

Official launch20 May 2026Source (opens in a new tab)
Technical report first posted18 May 2026Source (opens in a new tab)

Sources reviewed