Azure OpenAI gpt-audio model general availability

Question

Hi Team,

In our azure openai resource I see there is a model called "gpt-audio (version:2025-08-28)" that I have deployed and tested it out through the Chat playground. When I click "View Code" at the top left in the chat playground the code snippet shows me that the api_version is 2025-01-01-preview . Does this mean that this gpt-audio model is still in preview?

I can see in the documentation in the page here https://learn.microsoft.com/en-us/azure/ai-foundry/foundry-models/concepts/models-sold-directly-by-azure?pivots=azure-openai&tabs=global-standard-aoai%2Cstandard-audio%2Cglobal-standard#gpt-4o-audio-models there is no indication on this model being preview so wanted to confirm as I might be little confused with API version saying "preview" but the model name does not indicate anything.

Second question is that I don't fine any documentation around how to use this specific model as most documentation seems to be pointing towards using the gpt-realtime model. But nothing was found around the gpt-audio model usage so wanted some clarity on that.

Answer 1

Hello Akundy, Vyas,

Welcome to the Microsoft Q&A and thank you for posting your questions here.

I understand that you would like to know Azure OpenAI gpt-audio model general availability.

Your model is GA. The “preview” in api-version does not mean the model is preview. It only reflects the API surface you’re calling. For real-time voice, use gpt-realtime; for batch or async audio tasks, stick with gpt-audio via /chat/completions.

Regarding your questions:

Does api_version=2025-01-01-preview mean the gpt-audio model is still in preview?

No. The model lifecycle (GA vs preview) is separate from the API version name.

Why does the documentation page you linked not show a preview label?

That page Models sold directly by Azure - https://learn.microsoft.com/en-us/azure/ai-foundry/foundry-models/concepts/models-sold-directly-by-azure lists GA models without preview tags. This matches the Model Catalog, which shows gpt-audio as GA. The confusion arises because:

• Older audio models (e.g., gpt-4o-audio-preview) were explicitly labeled preview.
• The API version still uses -preview for audio completions features, even though the model is GA.

Why is there no dedicated documentation for gpt-audio like for gpt-realtime?

• gpt-audio is designed for asynchronous audio input/output (e.g., text-to-speech, audio summarization) using the Chat Completions API with audio modalities.
• The usage pattern is documented under Audio generation quickstart and Audio capabilities pages, not under a separate gpt-audio article. - https://learn.microsoft.com/en-us/azure/ai-foundry/openai/audio-completions-quickstart and https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/audio
• gpt-realtime has its own docs because it uses a different API (Realtime API) for low-latency, streaming voice interactions (WebRTC/WebSocket and https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/realtime-audio-webrtc

So, if you would like to look for gpt-audio usage use the below links:

• https://learn.microsoft.com/en-us/azure/ai-foundry/openai/audio-completions-quickstart
• https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/audio

I hope this is helpful! Do not hesitate to let me know if you have any other questions or clarifications.

---

Please don't forget to close up the thread here by upvoting and accept it as an answer if it is helpful.

Answer 2

Hello

It looks like you’re exploring the new Azure OpenAI gpt-audio model, and I can understand the confusion around its availability and documentation.

API Version & Availability: The API version 2025-01-01-preview indicates that while the gpt-audio model may be generally available (GA) in certain contexts, some features or endpoints may still be under preview. This is quite common during gradual rollouts, models are marked GA first , and supporting APIs/documentation catch up shortly after.

I recommend keeping an eye on the Azure OpenAI release notes and API version history for the most up-to-date GA status.

Documentation & Usage: The current documentation often points to gpt-realtime for examples. The gpt-audio model is newer, and Microsoft is progressively expanding its documentation.

In the meantime, you can explore this useful resource for guidance:

• Audio capabilities in Azure OpenAI in Azure AI Foundry Models

These cover the current supported features and SDK guidance for audio-related capabilities.

To help you better, could you share:

• Are you running into any specific implementation issues with gpt-audio?
• Have you checked whether your region supports this model yet?
• Would you like me to share a sample code snippet or use case example (e.g., transcription, audio generation, or real-time interaction)?

Hope this clears things up!