OpenAI has launched GPT-Live, a new generation of voice models designed to make conversations with ChatGPT feel faster, more natural and closer to speaking with another person.
Two GPT-Live Models Roll Out Globally
The launch includes two versions of the model.
Reuters reported that OpenAI will roll out GPT-Live-1 and GPT-Live-1 mini to users globally.
OpenAI said GPT-Live is rolling out across iOS, Android and ChatGPT.com, with GPT-Live-1 becoming the default ChatGPT Voice model for Go, Plus and Pro users and GPT-Live-1 mini becoming the default for Free users.
The company is also preparing developer access. OpenAI introduced three audio models for its developer platform in May to make voice-based software agents more conversational and capable of completing tasks in real time.
Voice AI That Can Listen While Talking
The biggest technical change is that GPT-Live does not have to wait for one person to fully finish before responding.
OpenAI said GPT-Live continuously processes input while generating output, allowing the model to decide many times per second whether to speak, listen, pause, interrupt or use a tool.
Cybernews reported that the system is designed to make chatbot conversations more realistic and closer to fluid dialogue.
That makes the voice experience less rigid than older systems.
OpenAI said earlier cascaded voice systems could lose information across separate speech-to-text, language and text-to-speech models, while turn-based models still had to wait for silence before responding.
Cybernews reported that GPT-Live can acknowledge users with brief phrases such as “mhm” or “I understand,” and can adjust when those responses happen depending on the flow of conversation.
GPT-5.5 Handles Harder Tasks in the Background
GPT-Live is also designed to keep talking while another model handles more complex work. GPT-Live can delegate questions requiring web search, deeper reasoning or more agentic capabilities to a model such as GPT-5.5, while continuing the conversation. GPT-Live can hand off complex tasks like web searches to background systems without interrupting the flow of dialogue.
GPT-Live will use GPT-5.5 in the background at launch and will be updated as newer frontier models are released. That structure points to a future where voice assistants do not only answer quick questions but also manage longer-running tasks while remaining conversational.
Safety Testing Expands for Voice Risks
OpenAI is also emphasizing safety because voice interactions can feel more personal than typed chats. OpenAI tested GPT-Live for safety risks including self-harm, psychosis, mania, emotional AI dependence, violence and sexual content.
The company says the system has built-in protections. GPT-Live can steer the model toward a safer response, surface safety messaging or resources, or end a voice conversation in higher-risk cases. The safety focus comes after public backlash and legal action against OpenAI over alleged chatbot harms.
A More Human Voice Assistant Race
The launch shows how quickly AI assistants are moving from text boxes to real-time conversation. More than 150 million people each week talk to ChatGPT using features like Voice and Dictation.
GPT-Live could make ChatGPT feel less like a tool waiting for prompts and more like a live assistant that can listen, respond, pause and continue working in the background. The bigger test will be whether users find that more natural, more useful or simply more unsettling when AI starts speaking back in real time.
