OpenAI introduces GPT-Live for ChatGPT Voice

On July 8, 2026, OpenAI introduced GPT-Live as a full-duplex voice model that can listen and speak simultaneously. It announced global rollouts of GPT-Live-1 and GPT-Live-1 mini. On July 31, it added SynthID watermarking to supported audio generated by ChatGPT Voice and the OpenAI API.

Visit original source