Skip to content
Model infoModelsSnapshot: 15 Oct 2026

StepFun / StepAudio

StepAudio 2.5 Realtime

Speech conversation with audio replies, turn detection and interruptible WebSocket interaction.

Speech & voice
availableClosed weightsRealtime audioSpeech generationAudio understanding
Official documentation (opens in a new tab)Reviewed 12 Oct 2026
4 of 16 specification fields collected
Context windowNot collectedAwaiting source review
Maximum outputNot collectedAwaiting source review
Input billingSee ratesStepFun China
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
Audio, TextSource for Input modalities (opens in a new tab)
Output modalities
Audio, TextSource for Output modalities (opens in a new tab)
Reasoning
Not collected
Tool calling
Not collected
Structured output
Not collected
Streaming
SupportedSource for Streaming (opens in a new tab)

Bidirectional WebSocket interaction.

Batch processing
Not collected
Fine-tuning
Not collected
Knowledge cutoff
Not collected
Release date
Not collected
Parameters
Not collected
Architecture
Not collected
Languages
Not collected
Context caching
SupportedSource for Context caching (opens in a new tab)

The regional pricing table explicitly publishes cached-input rates.

Access & pricing

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

StepFun China

Provider source (opens in a new tab)
API model ID
stepaudio-2.5-realtime
Access
StepFun account and regional API key required. Documented availability, not a paid inference verification.
Regions
China platform (.com); deployment geography and data residency are not established by the hostname.
API endpoint URL
WSS wss://api.stepfun.com/v1/realtimeSource for API endpoint URL (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 12 Oct 2026
CategoryUsageRateUnit
Speech conversation tokensInput1CN¥10.001 million tokens
Speech conversation tokensCached input1CN¥2.001 million tokens
Speech conversation tokensOutput1CN¥70.001 million tokens

Rate conditions

  1. Regional pay-as-you-go rate as reviewed on 2026-10-12. Input, cached input and output are distinct charges; no currency conversion. Subscription allowances are separate.
Pricing source (opens in a new tab)

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

StepFun International

Provider source (opens in a new tab)
API model ID
stepaudio-2.5-realtime
Access
StepFun account and regional API key required. Documented availability, not a paid inference verification.
Regions
International platform (.ai); deployment geography and data residency require account-specific confirmation.
API endpoint URL
WSS wss://api.stepfun.ai/v1/realtimeSource for API endpoint URL (opens in a new tab)
Languages
EnglishSource for Languages (opens in a new tab)

This endpoint-specific page documents English; do not inherit other generations’ language lists.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 12 Oct 2026
CategoryUsageRateUnit
Speech conversation tokensInput1$1.501 million tokens
Speech conversation tokensCached input1$0.301 million tokens
Speech conversation tokensOutput1$10.001 million tokens

Rate conditions

  1. Regional pay-as-you-go rate as reviewed on 2026-10-12. Input, cached input and output are distinct charges; no currency conversion. Subscription allowances are separate.
Pricing source (opens in a new tab)

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

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.

StepFun China

stepaudio-2.5-realtime
Documented
Support evidence (opens in a new tab)
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
Cached input is billed separately from uncached input and output; see endpoint rates.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Not collected
Cache reads, writes & storage
ChargeApplies toRate / unit
Cached input1Speech conversation tokensCN¥2.00 / 1 million tokensPricing evidence (opens in a new tab)

Rate conditions

  1. Regional pay-as-you-go rate as reviewed on 2026-10-12. Input, cached input and output are distinct charges; no currency conversion. Subscription allowances are separate.

Observed 12 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.

Caching reviewed 12 Oct 2026

StepFun International

stepaudio-2.5-realtime
Documented
Support evidence (opens in a new tab)
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
Cached input is billed separately from uncached input and output; see endpoint rates.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Not collected
Cache reads, writes & storage
ChargeApplies toRate / unit
Cached input1Speech conversation tokens$0.30 / 1 million tokensPricing evidence (opens in a new tab)

Rate conditions

  1. Regional pay-as-you-go rate as reviewed on 2026-10-12. Input, cached input and output are distinct charges; no currency conversion. Subscription allowances are separate.

Observed 12 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.

Caching reviewed 12 Oct 2026

Voice choices

7 voices

Choose a voice within StepAudio 2.5 Realtime. These are API voice identifiers, not additional model releases. Source links document each voice.

Published voice catalog
Voice / API IDLanguage & accentDelivery
Calm, gentle male voicesoft-spoken-gentlemanEnglishOfficial international system voice. Select before the session first produces audio.Voice source (opens in a new tab)
Deep, magnetic male voicemagnetic-voiced-maleEnglishOfficial international system voice. Select before the session first produces audio.Voice source (opens in a new tab)
Youthful, energetic voicevibrant-youthEnglishOfficial international system voice. Select before the session first produces audio.Voice source (opens in a new tab)
Bright, lively female voicelively-girlEnglishOfficial international system voice. Select before the session first produces audio.Voice source (opens in a new tab)
Light, breezy female voicelivelybreezy-femaleEnglishOfficial international system voice. Select before the session first produces audio.Voice source (opens in a new tab)
Elegant, gentle female voiceelegantgentle-femaleEnglishOfficial international system voice. Select before the session first produces audio.Voice source (opens in a new tab)
Confident male voicezixinnanshengEnglishOfficial international system voice. Select before the session first produces audio.Voice source (opens in a new tab)

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

Inferred for this named native commercial service from official model/API documentation and comparison with StepFun’s public model repositories. No release of weights for this exact named service was found in the reviewed inventory. Public Step 3/3.5/3.7, Step-Audio Mini/R1/R1.1, Step1X and StepVideo releases are excluded; this does not assert that every StepFun model is closed.

Classification evidence (opens in a new tab)

Sources reviewed