OpenAI on Wednesday launched GPT-Live, a new generation of voice AI models that is designed to enable more natural voice conversations. The company said GPT-Live is built on a full-duplex architecture, which means the AI can listen and speak at the same time instead of waiting for one person to finish before responding, according to the OpenAI official announcement.
GPT-5.6 launch gets approval
The GPT-Live announcement came a day after OpenAI said it would publicly release its flagship GPT-5.6 AI model following a temporary delay requested by the Trump administration over cybersecurity concerns.
Read here: Trump administration approves OpenAI’s GPT-5.6 rollout after safety testing – Here’s what’s next
OpenAI announced in a post on X late Tuesday that it would publicly release its newest and most advanced AI model series, GPT-5.6, on Thursday. OpenAI CEO Sam Altman also confirmed the launch in a separate post on X, writing, “happy building.” GPT-5.6 is OpenAI’s latest generation of AI models and includes three versions: Sol, the flagship model, along with Terra and Luna.
OpenAI had earlier planned to release GPT-5.6 publicly, but decided to delay the launch after a request from the Trump administration. According to OpenAI, the U.S. government asked the company to postpone the public release because of cybersecurity concerns related to advanced AI models.
How GPT-Live voice works
Following the GPT-5.6 rollout announcement, OpenAI said GPT-Live’s full-duplex system allows users and the AI to speak simultaneously during conversations. OpenAI said GPT-Live can show that it is listening by giving small responses such as “mhmm” or “yeah” while a person is speaking. The company added that GPT-Live can also take part in quick back-and-forth conversations without long pauses.
The company said that if a user asks a question that needs web search, deeper reasoning or more complex work, GPT-Live can hand that task to another advanced AI model working in the background. OpenAI said GPT-Live keeps talking with the user while the background model processes the request.
At launch, OpenAI said GPT-Live uses GPT-5.5 as the background model for these advanced tasks. The company added that GPT-Live will automatically use newer frontier AI models in the future as they are released.
GPT-Live rollout begins
OpenAI has started rolling out two versions of the new models: GPT-Live-1 and GPT-Live-1 mini. According to the company, both versions are being released to ChatGPT users around the world starting today. OpenAI also said it plans to bring GPT-Live to its API, allowing developers and businesses to build apps using the new voice technology.
