OpenAI is preparing a next-generation voice assistant and a consumer device

OpenAI is preparing to take a new step in its development with the launch of a next-generation audio model and its first voice-centered physical device. This strategy marks a turning point for the company, which aims to establish itself in the voice assistant space and reduce dependence on screens.

A next-generation voice model

OpenAI is currently working on an audio model that should significantly change the conversational experience with ChatGPT. Planned for the first quarter of 2026, this new system is based on a completely redesigned architecture that promises to correct the main weaknesses of current voice assistants.

The announced improvements are significant. The model will produce a more natural and expressive voice, far from the robotic intonations we are used to. But the real innovation lies in conversation management: the system will be able to speak at the same time as the user, handle interruptions and overlapping exchanges. These capabilities will finally allow dialogues that resemble genuine human conversations, where you can cut each other off or instantly jump back into a topic.

This technological development is far from insignificant. Today, voice assistants like Alexa or Siri force the user to wait for the end of the response before they can intervene, which creates an artificial latency in the exchange. OpenAI's new model aims to remove this barrier to make interaction smoother and more intuitive.

A strategic internal reorganization

To carry out this transformation, OpenAI has undertaken a major restructuring of its teams over the past two months. The engineering, product and research departments have been merged into a single entity dedicated to the development of audio technologies.

This organizational change is evidence of a shift in priorities within the company. While OpenAI made its name with textual and visual models, voice will now become a full strategic pillar. This consolidation allows faster development in an area where the company had been somewhat behind compared with its other generative AI capabilities.

An audio-centered device ecosystem

OpenAI's first physical device is not expected to arrive for at least a year, which places its launch between late 2026 and early 2027But it will only be the first member of a larger family of interconnected products.

The approach envisaged by OpenAI is the following: create an ecosystem entirely centered on voice, where screens take a back seatAmong the formats explored internally are connected glasses, similar to what Meta offers with its smart Ray-Bans, as well as smart speakers without a screen. The goal is to design ambient AI companions rather than traditional tech gadgets.

This vision is part of a broader trend in the tech industry, where many players are trying to reinvent our relationship with technology. Meta is already developing glasses that can enhance hearing in noisy environments thanks to a system of five directional microphones. Google has been experimenting since June 2025 with the Audio Overviews, which turn search results into conversational summaries. Tesla integrates xAI's Grok chatbot into its vehicles to create a voice assistant capable of handling navigation and vehicle controls through natural dialogue. In short, everyone is launching their own little revolution.

Jony Ive's influence on the project

The acquisition of io Products in May 2025 for about $6.5 billion placed Jony Ive, the former chief designer at Apple, at the heart of OpenAI's hardware project. This collaboration is not just a PR stunt: it reflects a precise design philosophy.

Ive made reducing dependence on screens his priority. For him, audio-first devices represent an opportunity to "fix the mistakes" of recent consumer gadgets, often accused of being intrusive and promoting addiction. This approach aims to create a a new, more discreet and less intrusive computing era, where technology fades away to make room for the experience.

The challenges of an already crowded market

OpenAI is not the first to explore this territory. Many startups have tried their luck with mixed results. The Humane AI Pin, a screenless wearable device, burned through hundreds of millions of dollars before becoming a spectacular failure, bought by HP in February 2025 for its technological assets. The Friend AI, a pendant that records your daily life to offer companionship, raised as many privacy concerns as existential questions.

Other players nonetheless persist. Sandbar and a startup founded by Eric Migicovsky, creator of the Pebble watch, are developing connected rings that allow voice control of AI. These products are also expected to reach the market in 2026.

Audio as the interface of the future

All these efforts converge toward the same conviction: making audio the primary interface for interactions with technology. Smart speakers are already present in more than a third of American households, proof that users are ready to adopt voice interfaces for their daily tasks.

With its expertise in generative AI and Jony Ive's industrial know-how, OpenAI has serious assets to succeed where others have failed. The timing is also favorable: conversational AI models have reached sufficient maturity to deliver genuinely useful experiences, unlike earlier attempts that relied on still-immature technologies.

If OpenAI manages to create a truly natural voice assistant and elegantly designed devices, the company could unsettle established players like Apple, Google, or Amazon on their own turf. The battle of voice assistants may only be beginning!

The article "OpenAI is preparing a next-generation voice assistant and a consumer device" was published on the site Abondance.