OpenAI Launches Natural Full-Duplex Voice Models

OpenAI released two new conversational voice models, GPT-Live-1 and GPT-Live-1 mini, designed for more natural conversations with better turn-taking and interruption handling. These full-duplex models can speak and listen simultaneously, enabling features like live translation and visual responses. GPT-Live-1 mini will replace ChatGPT's current Advanced Voice Mode by default, with the larger model available to paid users. OpenAI envisions voice becoming the primary computing interface for complex tasks, noting over 150 million people already use ChatGPT voice features. However, the launch showed limitations, including unnatural-sounding Hindi translation with a heavy American accent. The company emphasized the feature is not designed as an AI companion and includes safeguards for teens and sensitive topics.