StepAudio 3 Chat | EmpirioLabs AI Docs

StepFun · Text Generation

POST /v1/chat/completions

Audio and text conversation model that reasons before answering, with function calling and JSON output.

At a glance

Field	Value
Model id	`stepaudio-3-chat`
Model release date	2026-09-09
Input modalities	Text, Audio
Output modalities	Text
Context window	256K
Weight precision	-
Max output tokens	131,072
Region	International
Features	reasoning, audio_in, function_calling, structured_output, multimodal
Native inference	No
New	Yes
Structured output	JSON Mode
Supported endpoints	`POST /v1/chat/completions`
Alternate model ids	`stepaudio-3-chat-preview`, `stepfun/stepaudio-3-chat`

Pricing

Charge	Spec	Rate
Input	per 1M prompt tokens	$1.50
Output	per 1M generated tokens	$3.50
Implicit cache read	per 1M cached input tokens	$0.30
Web Search (Linkup)	per call when invoked	$0.013

Example request

curl https://api.empiriolabs.ai/v1/chat/completions \
  -H 'Authorization: Bearer $EMPIRIOLABS_API_KEY' \
  -H 'Content-Type: application/json' \
  -d '{"model": "stepaudio-3-chat", "messages": [{"role":"user","content":"Hello"}]}'

Parameters

Parameter	Type	Required	Default	Description
`temperature`	number	no	`0.5`	Sampling temperature. · Range: 0 – 2
`top_p`	number	no	`0.9`	Nucleus sampling probability mass. · Range: 0 – 1
`max_tokens`	integer	no	`4096`	Maximum output tokens. Reasoning tokens count toward this limit, so a small budget can be spent entirely on thinking and return an empty answer. Leave unset for a sensible default, or allow at least a few thousand tokens. · Range: 1 – 131072
`stop`	array	no	-	Stop sequences.
`frequency_penalty`	number	no	`0`	Penalty for repeated tokens. · Range: 0 – 1
`modalities`	array	no	`["text"]`	Response modalities. Text output is currently exposed.
`response_format`	object	no	-	OpenAI-compatible response format. Use {“type”:“json_object”} for JSON object mode.
`reasoning_format`	enum	no	`"general"`	Reasoning trace format returned by StepFun. · Allowed: `general`, `deepseek-style`
`tools`	array	no	-	OpenAI-compatible function tools.
`tool_choice`	string	no	-	OpenAI-compatible tool choice.
`web_search_linkup`	boolean	no	false	Optional web search powered by Linkup. When enabled, recent web sources are retrieved using your latest user message as the query and provided to the model as additional context. Adds $0.013 per call when invoked on top of the model’s normal token cost. Disabled by default.
`disable_formatting`	boolean	no	false	When enabled, the gateway will not append the “Sources” footer to assistant responses that used Linkup web search. Useful when the model output is piped to another system that expects no decoration.

Notes

Accepts text and audio input and replies with text. The model reasons before answering and returns its reasoning alongside the answer, and those reasoning tokens bill as output tokens. Because reasoning consumes the output budget, a small max_tokens can be spent entirely on thinking and return an empty answer; leave max_tokens unset for a sensible default, or allow at least a few thousand tokens. Function calling and JSON object output are supported. Linkup web search is available for text prompts. This model is in preview and its capabilities may change.

Machine-readable schema: GET https://api.empiriolabs.ai/v1/models/stepaudio-3-chat.