OpenAI has released GPT-Live-1 and GPT-Live-1 mini, a new generation of voice models designed to make ChatGPT conversations feel more natural. The models are full-duplex, meaning they can listen and speak at the same time, so users can interrupt, pause and resume more naturally during a conversation.
GPT-Live is rolling out globally in ChatGPT on iOS, Android and the web. GPT-Live-1 will become the default voice model for Go, Plus and Pro users, while GPT-Live-1 mini will power the default voice experience for Free users.
The update replaces the older Advanced Voice setup, which used separate systems for speech-to-text, language reasoning and text-to-speech. OpenAI says the new models are better at deciding when to respond, when to stay silent and how to handle interruptions.
Voice Mode Becomes More Conversational
OpenAI said the new models address common issues in voice assistants, including interrupting users too early or failing to understand more complex questions. GPT-Live can continue listening while a person speaks and can wait through long pauses without forcing a response.
The models can also support live translation and longer spoken conversations. Atty Eleti, ChatGPT Voice’s product lead, said he has used the feature for 30- to 40-minute conversations while walking.
When a user asks a more complex question, GPT-Live can send the request to newer text models such as GPT-5.5 for search, reasoning or agentic work while keeping the voice conversation going. OpenAI says the new voice mode can also show visual results through supported widgets when useful.
The company’s ChatGPT Voice help page says Live can use web search and memory, work with text and images in the same chat, and display visual results. It does not initially support video, screen sharing, connected apps or plugins.
OpenAI Sees Voice as a Future Computing Interface
OpenAI said more than 150 million people use ChatGPT through Voice and Dictation. The company sees spoken interaction as a potential primary interface for computing, especially as AI agents begin handling longer and more complex work.
Eleti said voice could eventually become the way people manage tasks now performed through Codex and ChatGPT. OpenAI has not announced any hardware tied to GPT-Live, though reports have suggested the company has explored AI-enabled earbuds.
Other companies are moving in the same direction. Apple and Amazon have been updating their assistants to handle more context and sound more conversational, while startups such as Sesame are building AI assistants designed for natural speech and background task completion.
OpenAI said GPT-Live includes safeguards for teen users and crisis-related conversations, including age-appropriate responses and resources for topics such as self-harm. The company also said the system is not designed to act as an AI companion.
The new voice mode still has limitations. During a demo, its Hindi live translation sounded heavily accented and somewhat unnatural, according to TechCrunch. OpenAI said GPT-Live is optimized for most spoken languages, but did not specify which languages perform best.
Featured image credits: Roboflow Universe – YOLO
For more stories like it, click the +Follow button at the top of this page to follow us.