OpenAI Introduces New GPT-Live Voice Models: Conversations Are Now More Natural

OpenAI, a global leader in AI, has announced the GPT-Live-1 and GPT-Live-1 mini models, taking real-time interaction with users to a new level. This technology makes the process of talking to ChatGPT as close as possible to natural human conversation. The new models not only respond faster but also have the ability to directly understand and adapt to the speaker's speech. This is reported by Techcrunch.com reports.

The most important feature of the new generation models is their operation in a "full-duplex" system. This means that ChatGPT can now listen and speak at the same time. While previous versions required waiting for the user to finish speaking, GPT-Live models allow for natural interruptions or the use of live translation functions. According to Ixbt.com, the company is migrating the Advanced Voice Mode feature in the ChatGPT app entirely to the GPT-Live-1 mini model.

Technological revolution and intellectual potential

Previously, voice interaction was a complex three-step process: speech was converted to text, a text model prepared a response, and then it was converted back to voice. The new models shorten this chain and process information directly. This eliminates problems such as the system pausing inappropriately while the user is speaking or failing to provide sufficiently intelligent answers. Additionally, the new voice mode can connect to the latest text models like GPT-5.5 to perform complex logical tasks.

According to OpenAI representatives, GPT-Live models have the ability to remain silent for long periods, grasp the context of the conversation, and engage only when necessary. Furthermore, the new system can also provide visual information. This allows users not only to talk to ChatGPT but also to work with interactive visual materials. This approach is also being actively used by startups like Monogram.

Voice interface — the foundation of the future

Atti Eleti, head of the ChatGPT Voice project, noted that voice interaction could become the primary way of working with computers in the future. He shared his experience of conducting long, meaningful conversations lasting up to 40 minutes using this feature. The company sees the voice assistant not just as a search tool, but as a primary interface for managing complex tasks.

Currently, over 150 million people use ChatGPT's voice and dictation features. In a competitive environment, Apple and Amazon are also working on making their virtual assistants more natural. OpenAI is reportedly considering not just software, but also future hardware like dedicated AI-earbuds, though no information about hardware was provided at the official launch.