OpenAI has announced the launch of GPT-Live voice model in ChatGPT. The company describes it as “a new generation of voice models that make talking with AI feel much more like having a real conversation.” The company confirmed that “GPT‑Live is built on a full-duplex architecture, meaning it can listen and speak at the same time.” Here’s everything to knoww about the new AI voice model for ChatGPT.
GPT-Live in ChatGPT: Everything to Know
GPT-Live is OpenAI’s new full-duplex voice model designed to make conversations with ChatGPT feel more natural. Unlike previous voice systems that waited for users to finish speaking before responding, GPT-Live can listen and speak simultaneously, allowing for more fluid conversations with natural acknowledgements, quicker back-and-forth exchanges, and fewer awkward interruptions.
The new voice model also brings improved intelligence by working alongside OpenAI’s latest frontier models. When a conversation requires web search, deeper reasoning or more complex problem-solving, GPT-Live can seamlessly hand off the task to GPT-5.5 in the background while continuing the conversation. OpenAI says this architecture allows users to keep talking without waiting for the AI to finish processing complex requests, with future frontier models also expected to power GPT-Live over time.
According to OpenAI, GPT-Live represents a significant shift from previous voice architectures. Earlier versions relied on separate speech-to-text, language and text-to-speech models, which often resulted in slower responses and information loss between stages. Even Advanced Voice Mode, despite reducing latency, still relied on turn-based conversations that could interrupt users after brief pauses. GPT-Live replaces this with continuous processing, allowing it to decide in real time whether to respond, continue listening, pause or use external tools.
OpenAI says the new architecture also improves performance in real-world tasks. Internal testing showed GPT-Live outperforming Advanced Voice Mode in areas such as scientific reasoning, web search and multi-turn voice support scenarios. The company also reports that users preferred GPT-Live for conversational flow, naturalness and overall interaction quality during human evaluations.
Also Read: OpenAI Rolls Out Codex in ChatGPT Mobile App
The updated ChatGPT Voice experience includes several practical improvements. Users can interrupt ChatGPT naturally, pause to think without being cut off, or ask it to remain silent while listening. The model is also better at filtering out background noise and can display visual cards for information such as weather, stocks and sports while voice conversations are ongoing. OpenAI has also refreshed all nine built-in ChatGPT voices specifically for GPT-Live.
On the safety front, OpenAI says GPT-Live includes dedicated safeguards designed specifically for voice interactions. The model has undergone expanded audio-focused safety testing covering areas such as self-harm, emotional reliance, violence and inappropriate content. It can intervene during conversations if potentially unsafe responses are detected, while additional protections have been added for teen users through parental controls and age-appropriate behaviour. The company also notes that GPT-Live only supports predefined voices and includes safeguards to prevent impersonation of real people.
GPT-Live is rolling out globally starting today across ChatGPT on iOS, Android and the web. GPT-Live-1 will become the default voice model for Go, Plus and Pro subscribers, while Free users will receive GPT-Live-1 mini. At launch, voice conversations with video and screen sharing are not supported in GPT-Live, although OpenAI says those capabilities are planned for a future update. Moreover, one can still access legacy versions of ChatGPT Voice, including Standard and Advanced Voice Mode, where these features are available.
Author: The Mobile Indian Network
Source: The Mobile Indian
Reviewed By: Editorial Team
