wanton pulsar
With the advent of GPT-5, Custom GPTs now support Advanced Voice Mode. However, when this mode is enabled, they become significantly less capable of following complex instructions defined in the Custom GPT configuration — including code execution and other advanced behaviors.
It is therefore necessary for OpenAI to introduce a configuration flag in Custom GPT settings called “Advanced Voice Mode”. This flag could be ON by default, but the author of the custom-gpt should have the option to switch it OFF to disable advanced voice mode and restore the higher-level reasoning and capabilities of the standard model.
This behavior could be entirely analogous to the current “Recommended Model” selector, where the Custom GPT can suggest a default model to use for optimal results.
I would like OpenAI to consider implementing this small yet crucial improvement, so that more complex Custom GPTs powered by GPT-5 can be fully utilized with voice interaction.
What do you think? I’d love to hear your thoughts... Thank you!