OpenAI today released new conversational models called GPT-Live-1 and GPT-Live-1 mini, claiming they sound more natural and handle turn-taking better. These are full-duplex models, meaning they can talk and listen at the same time, allowing users to interrupt conversations naturally and include features such as direct translation.
The company is also replacing the current enhanced voice mode in ChatGPT with GPT-Live-1 mini by default. Paid tiers will be able to access the larger GPT-Live-1 model. The previous model combined a speech-to-text model to transcribe speech, a larger language model to generate responses, and a text-to-speech model to provide the final answer.
The company said at a press briefing that the new models address issues such as users interrupting during calls and not being smart enough to answer questions. New OpenAI models will query their latest text models, such as GPT-5.5, for search, reasoning, or agent capabilities while continuing the conversation.
OpenAI also showed that the model can remain silent for a long time and absorb the context of the conversation until it is addressed. Additionally, since the new voice mode has access to newer GPT models, it can also present some information in a visual format. Other startups like Monogram, which has raised $40 million in seed funding from DST and Lux Capital, are also relying on visual responses to make assistants more interactive.
The company said the new voice mode in ChatGPT is designed for longer conversations. During the briefing, ChatGPT Voice product manager Atti Eleti said that while walking, he talked to the voice function for 30-40 minutes.
OpenAI believes that voice can become the primary computing interface for performing complex work. Reports suggest that the company may launch a pair of headphones with artificial intelligence capabilities this year. However, it did not provide any information about the hardware products.
“We think that over time this will also open up the ability to use voice as sort of the primary interface for computing and managing increasingly complex, long-running agent work. We think people using Codex and ChatGPT are achieving some amazing use cases, and we think voice could be the future interface for all kinds of work,” Eleti said.
Over the past few years, OpenAI has been working on expanding the voice functionality to make ChatGPT’s voice mode sound more natural. The company said more than 150 million people communicate with ChatGPT using features such as voice and dictation.
Rivals are also trying to make their assistants more expressive.
Both Apple and Amazon have updated their assistants to be more conversational and better at processing context. Startups like Sesame, founded by Oculus co-founders Brendan Iribe and Ankit Kumar, have also launched AI assistants with more natural communication and task completion in the background.
OpenAI is moving in a similar direction, aiming to allow users to communicate with their assistant hands-free for longer. Despite the claim that the new voice mode sounds more natural, the company stressed that it is not aiming to make it an AI assistant. He noted that the new models have built-in safety measures to give teens age-appropriate answers and resources when the conversation turns to topics like self-harm.
The new voice mode still needs some work. During a demo where the company was demonstrating the Hindi live translation feature, the assistant had a thick American accent and spoke in Hindi that sounded stilted and had a slightly bookish tone. The company said the new mode is optimized for “the most common languages,” but did not specify which ones.
When you make purchases through links in our articles, we may receive a small commission. This does not affect our editorial independence.