OpenAI gives ChatGPT a new voice
OpenAI, an American company that makes AI programs, has a new way for people to talk with ChatGPT. On July 8, the company said it has started to offer two new voice models. One is called GPT-Live-1. The smaller one is called GPT-Live-1 mini.
With the new models, the AI can listen and speak at the same time. So a person and the AI can keep talking without stopping. When a question is hard, the voice model can pass it to a bigger model called GPT-5.5. The conversation can go on while GPT-5.5 works on the answer.
Users can choose how deep they want the answers to be. There are three levels: Instant, Medium and High. For some topics, like the weather or sports, ChatGPT shows the information on a visual card on the screen.
GPT-Live-1 is now the standard voice model for people who pay for Go, Plus or Pro. GPT-Live-1 mini is the standard model for people who use ChatGPT for free. The new models are coming to users around the world step by step, on iOS and Android phones and on the ChatGPT website.
OpenAI also tested how safe the new models are. It looked at difficult areas such as self-harm, signs of serious mental illness, violence and sexual content. The company says the new models did as well as the older Advanced Voice Mode, or better. At the start, the new voice models cannot be used together with video or screen sharing.
Glossary
- 246 words
- 18 sentences
- 3 mins to read




