Field operator providing voice input
AI Models & Assistants / Knowledge library
GPT-Audio
For audio inputs and outputs with Chat Completions API
Why learn GPT-Audio?
Offers the documented OpenAI model capabilities through hosted.
The problem it solves
Provides model inference for supported text / audio inputs.
Where it came from
GPT-Audio: For audio inputs and outputs with Chat Completions API Published aliases: gpt-audio-2025-08-28. Input: text, audio. Output: text, audio. Checked 2026-10-13. Model capabilities are distinct from tools available in a particular agent.
Build your understanding
Work through these exercises in order. Try the task before opening a reference.
- Explain what GPT-Audio does.
- Identify a use case for GPT-Audio.
- Describe a limitation of GPT-Audio.
- Name a compatible tool for GPT-Audio.
- Compare GPT-Audio against an alternative.
Documentation and sources
- Primary source / model card (developers.openai.com)
Catalog research notes
- See the AI Workbench for source dates, pricing evidence and configuration details.
Connect the ideas
Built on
Real World Example
Voice-Activated Field Maintenance Assistant
Industrial field technicians who need hands-free access to technical manuals and diagnostic guidance while performing complex equipment repairs.
The business problem
Technicians often wear protective gloves and operate in environments where typing on a mobile device is physically impossible or hazardous, leading to delays in accessing critical repair documentation.
How this topic solves it
The gpt-audio model allows technicians to interact with a centralized knowledge base using natural voice commands, receiving direct audio responses that guide them through repair steps without requiring manual input.
- Capture raw audio input from the technician using a mobile microphone interface.
- Send the audio data directly to the v1/chat/completions endpoint using the gpt-audio model ID.
- Configure the request to accept audio modalities for both input and output.
- Receive the synthesized audio response from the model and stream it directly to the technician's headset.
- Utilize function calling to retrieve specific equipment schematics if the model requires external data context.
The flow shows how audio data moves from the field technician to the model and back as a response.
Captures and transmits audio
Processes audio and generates response
Plays audio output to technician
- Technician → speaks query → Mobile App
- Mobile App → sends audio to v1/chat/completions → gpt-audio Model
- gpt-audio Model → streams audio response → Headset
Expected business outcome
Technicians achieve higher operational efficiency and safety by maintaining focus on the equipment while receiving real-time, hands-free technical support.
Limitations and checks
The model processes audio as tokens, which can lead to latency during complex reasoning tasks; developers should monitor the response time to ensure it meets the requirements for real-time field interaction.