OpenAI introduces GPT-Live voice models
GPT-Live makes ChatGPT voice conversations feel more natural, fast, and fluid. Get smoother back-and-forth, fewer pauses, and better interruptions across devices.
OpenAI has announced GPTLive, a new set of voice models designed to make conversations with ChatGPT feel more natural and responsive. The company says the models can listen and speak at the same time, allowing for smoother backandforth, fewer awkward pauses, and better handling of interruptions or moments when a user needs time to think.
At launch, GPTLive will roll out to ChatGPT users globally on iOS, Android, and the web. OpenAI said GPTLive1 and GPTLive1 mini will power the Voice experience, with the more capable version using GPT5.5 in the background for tasks that require deeper reasoning, search, or more complex work. The company also plans to bring the models to its API soon.
OpenAI said the update is intended to improve voice use cases such as handsfree help, language practice, translation, and other everyday conversations. The company added that the system includes new safety measures for voice interactions, along with controls for teen users and protections aimed at reducing harmful or inappropriate responses.