I would like to be able to reference internal endpoints with Azure OpenAI, AWS Bedrock, and/or models hosted on an internal server.
Completed
Feb 1, 2026
Right now, the custom Voice Models feature has quite a few limitations. For example, if the model is not literally named gpt-4o-transcribe, it simply doesn’t appear in the Custom Voice Model list at all, even if it’s an audio transcription model.
Right now, the custom Voice Models feature has quite a few limitations. For example, if the model is not literally named gpt-4o-transcribe, it simply doesn’t appear in the Custom Voice Model list at all, even if it’s an audio transcription model.