Step-Audio 2 Mini
Step-Audio 2 Mini is the downloadable dialogue member of the Step-Audio 2 family. Its model card describes tool calls and retrieval that can use both text and acoustic material.
Retrieve sound as well as text
The card describes multimodal retrieval for real-world knowledge. This gives the family a route to use an audio reference rather than reducing every source to written words.StepFun Step-Audio-2-mini model card
Measure the full audio path
Practical guidance. When comparing self-hosting with the API, include audio encoding and waveform generation in latency. A language-only token rate does not describe the delay that a person hears.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API, Downloadable |
| Model identifiers | Step-Audio-2-ministep-audio-2-mini |
| License | Apache 2.0 for the model and code.StepFun Step-Audio-2-mini model card |
| Hosted option | The Realtime API also lists step-audio-2-mini.StepFun Realtime API model support |
| Acoustic retrieval | The model can change voice based on retrieved speech.StepFun Step-Audio-2-mini model card |
Related models
- Step-Audio 2
StepFun · StepAudio 2
Step-Audio 2 understands speech, music, environmental sound and emotion, with tools and search.
Launch date not confirmedAPI · Provider product - Step-Audio 2 Mini Base
StepFun · StepAudio 2
A base audio-language checkpoint in the Step-Audio 2 framework before dialogue tuning.
Launch date not confirmedResearch releaseDownloadable - Step-Audio 2 Mini Think
StepFun · StepAudio 2
A Step-Audio 2 Mini checkpoint for audio understanding and spoken dialogue.
Launch date not confirmedResearch releaseDownloadable