OpenAI adds prompt-to-voice and new voices to GPT-Live

OpenAI's API changelog lists two updates to GPT-Live, its real-time voice API.

On 15 October 2026, OpenAI said GPT-Live (gpt-live-1 and gpt-6-luna) gained a native integration with Decisions, so apps can act while the user is still speaking.

On 16 October 2026, OpenAI made prompt-to-voice available to all API customers. Developers can create a custom voice from a text description through the API or the voice Playground. OpenAI says creating a voice is free, while speech is billed at normal GPT-Live session rates. It also added nine voices covering German, French, Hindi, Japanese and Korean.

The full entries are in OpenAI's changelog.