OpenAI releases new voice models with full-duplex conversation for ChatGPT

OpenAI has released a new generation of voice models that bring full-duplex communication—the ability to speak and listen at the same time—to ChatGPT. The update, announced on July 8, marks a significant step toward making voice a primary interface for interacting with AI.

According to TechCrunch, OpenAI said in a press briefing that the new models solve two persistent issues: interrupting users while they are talking, and lacking sufficient intelligence to answer complex questions. The model can now stay silent for extended periods, absorbing conversation context until it is called upon to respond.

The company is replacing its current Advanced Voice Mode in ChatGPT with GPT-Live-1 mini by default. Users on paid tiers will be able to access the larger GPT-Live-1 model. The new voice mode is designed for longer conversations—ChatGPT Voice's product lead, Atty Eleti, said he has had 30- to 40-minute-long conversations with the model.

OpenAI believes voice could become the primary interface to computing, particularly for managing increasingly complex long-running agentic work. The company said that more than 150 million people talk to ChatGPT using voice today, and the new model is positioned to accelerate that adoption.

However, the new voice mode still needs work. During a demo of the live translation feature in Hindi, the assistant had a heavy American accent and spoke in Hindi that sounded unnatural, highlighting that cross-language realism remains a challenge.

Competitors are also moving quickly. Both Apple and Amazon have updated their assistants to be more conversational with better context handling. Startups like Sesame, founded by Oculus co-founder Brendan Iribe and Ankit Kumar, have also launched more expressive voice AI products. OpenAI's new voice models aim to differentiate through full-duplex capability and deeper contextual understanding.