OpenAI rolls out GPT-Live voice for ChatGPT on web and mobile

OpenAI launched GPT-Live, a new family of voice models for ChatGPT Voice, rolling out globally starting July 8, 2026. The full-duplex architecture enables simultaneous listening and speaking, handling interruptions and background tasks, replacing earlier voice systems. GPT-Live-1 becomes default for Go, Plus, and Pro users, while GPT-Live-1 mini serves Free users on iOS, Android, and web, with API access planned soon.

OpenAI has launched GPT-Live, a new family of voice models for ChatGPT Voice, with GPT-Live-1 and GPT-Live-1 mini rolling out globally starting July 8, 2026. The update moves ChatGPT Voice beyond turn-based conversations by using a full-duplex architecture https://www.testingcatalog.com/openai-prepares-bidirectional-voice-mode-for-rollout-on-chatgpt/ , allowing the system to listen and speak simultaneously, handle interruptions, pause when the user is thinking, and continue a conversation while deeper tasks run in the background. The launch targets everyday ChatGPT users first. GPT-Live-1 becomes the default voice model for Go, Plus, and Pro users, while GPT-Live-1 mini becomes the default for Free users. The rollout covers ChatGPT on iOS, Android, and web, while Business, Enterprise, and Edu workspaces are not included at launch. OpenAI says API access is planned soon, with developers and enterprises able to register interest. GPT-Live can delegate harder requests to frontier models behind the scenes. At launch, that means GPT-5.5 handles search, reasoning, or more complex work while GPT-Live keeps the spoken conversation active. Users can also select Instant, Medium, or High intelligence levels where available, trading response speed for deeper reasoning. ChatGPT Voice now supports visual cards for topics such as weather, stocks, and sports, as well as web search, memory, text, and images in supported accounts. This is a direct replacement path for OpenAI’s earlier voice systems. The original Standard Voice Mode chained speech-to-text, an LLM, and text-to-speech, which added latency. Advanced Voice Mode processed audio using a single model but still relied on discrete turns. GPT-Live changes that model by continuously processing input and output, allowing it to decide many times per second whether to speak, listen, pause, interrupt, or call a tool. The initial release has clear limits. Live does not support video or screen sharing at launch, and it is not initially available in Temporary Chats, the ChatGPT desktop app, Work, Codex, or custom GPTs. Advanced Voice Mode remains available for eligible users who need mobile video or screen sharing. OpenAI https://www.testingcatalog.com/tag/chatgpt/ is positioning GPT-Live as the foundation for longer, more agentic voice work, not just casual chat. The company says more than 150 million people use ChatGPT Voice and Dictation each week, giving this rollout immediate scale across consumer ChatGPT.