Azure OpenAI Realtime API

Artur Dereń 0 Reputation points
2025-08-07T16:51:06.7866667+00:00

In Azure Foundry, the model "gpt-4o-realtime-preview-2025-06-03" is marked as:

"Legacy model version: This deployment is using legacy model gpt-4o-realtime-preview version 2025-06-03 which will be retired on 9/01/2025 local time. When the version retires, inferencing will return error responses and all deployments for this version will not be supported. To avoid disruptions, you can edit the deployment now to update the deployment to a supported version, if a new version is available."

However, there is currently no newer version available on the list.

Will a new model be released in the coming days?

We’ve also noticed that this version does not support all features available in the same version of the model accessed directly via the OpenAI API (e.g., more voices, speech speed control).

Will these capabilities also be added soon?

Azure AI Speech
Azure AI Speech
An Azure service that integrates speech processing into apps and services.
0 comments No comments
{count} votes

2 answers

Sort by: Most helpful
  1. Artur Dereń 0 Reputation points
    2025-08-07T16:52:28.8066667+00:00

    User's image

    User's image

    0 comments No comments

  2. Pavankumar Purilla 10,350 Reputation points Microsoft External Staff Moderator
    2025-08-08T07:07:36.3066667+00:00

    Hi Artur Dereń,
    Currently, Azure OpenAI’s gpt-4o-realtime-preview-2025-06-03 model is marked as a legacy version scheduled for retirement on September 1, 2025, after which all deployments using it will return errors. Although the portal warns to upgrade, there is no newer version available in the dropdown at this time, and Microsoft has not officially announced a replacement release date. Historically, Azure releases updated snapshots of preview models shortly before older ones retire, so it is likely a new version will appear closer to the retirement date. The most recent “What’s New” updates for Azure OpenAI introduced additional voices and improvements in real-time audio streaming performance, but the current Azure version of this model still lacks some features available in OpenAI’s direct API—such as speech speed control and a wider set of audio configuration options. Azure has been gradually closing this gap in newer preview releases, so feature parity may improve in upcoming versions, though no public roadmap confirms when.


Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.