Tasks and inputs
StepAudio 3 Chat Preview is listed for Audio-text-to-text. It accepts text, audio and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
SyncDev model directory · Profile updated Oct 8, 2026
StepAudio 3 Chat Preview understands speech and text through Chat Completions and returns text replies with tool support.
The specification score summarizes recorded features and limits. It is not a benchmark result.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
StepAudio 3 Chat Preview understands speech and text through Chat Completions and returns text replies with tool support.
StepAudio 3 Chat is the Audio 3 route for an application that submits speech or text one turn at a time and receives text. It supports context understanding and tool calling. It is useful when a product needs to interpret a recording or spoken request but already has its own display or speech synthesizer.
The application maintains its message history for Chat Completions. Automatic turn management, voice activity detection and continuous bidirectional audio belong to the separate Realtime API. Selecting a speech-understanding model does not automatically add a complete voice conversation system to an ordinary HTTP client.
The documented trial identifier is stepaudio-3-chat-preview. StepFun describes trial access as temporary and expects a formal paid version to replace it. This listing leaves long-term prices and undocumented context limits unknown. Check the account’s enabled models before relying on the preview in a service.
Send the same request as typed text and as recorded speech. Include a proper name, a number and a correction made mid-sentence. Check whether the response follows the final intent and whether a tool call receives the correct arguments. Keep the original audio with the returned text so transcription or interpretation errors can be separated from errors in the downstream application.
Compare the model on the work you need to complete.
StepAudio 3 Chat Preview is listed for Audio-text-to-text. It accepts text, audio and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
Measure a completed task using the prompts, documents and tools your application will use. Include reasoning and retries in the result.
Use the model identifier and account requirements for the selected API route. Confirm its request limits and output format with one complete task before connecting a larger workflow.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
ACE-Step
ACE-Step/acestep-transcriber is a source-linked Hugging Face repository assigned to Audio-text-to-text. Its recorded task interface accepts audio, text and produces text.
ai-babai
Hugging Face repository for Audio-text-to-text.
bartowski
Hugging Face repository for Audio-text-to-text.
bartowski
bartowski/mistralai_Voxtral-Small-24B-2507-GGUF is a source-linked Hugging Face repository assigned to Audio-text-to-text. Its recorded task interface accepts audio, text and produces text.
cstr
Hugging Face repository for Audio-text-to-text.
DeSTA-ntu
Hugging Face repository for Audio-text-to-text.