OpenAI consolidates teams to push audio-first products and new voice model

OpenAI has combined multiple engineering, product and research teams over the past two months to overhaul its audio models as it prepares an audio-first personal device expected in about a year, The Information reports. The company’s next audio model, slated for early 2026, is said to aim for more natural speech, better handling of interruptions and the ability to speak while a user is talking — capabilities current systems lack.

OpenAI also envisions a family of companion-like devices, possibly including glasses or screenless smart speakers. The move reflects a broader industry shift toward audio interfaces. Smart speakers are already in more than a third of U.S. homes. Meta added a directional listening feature to its Ray-Ban glasses, Google has experimented with "Audio Overviews" that turn search results into conversational summaries, and Tesla is integrating xAI’s Grok chatbot into vehicles to provide voice control for navigation and climate.

Startups and product failures underline both the opportunity and the risk. The Humane AI Pin became a cautionary example after heavy spending, Friend AI’s pendant has raised privacy concerns, and several companies — including Sandbar and a venture led by Pebble founder Eric Migicovsky — are developing AI rings due in 2026 to enable gesture and voice interactions.

OpenAI’s hardware push has design ambitions tied to reducing device addiction.

Key Topics

AI, Tech, Audio Ai, Openai, Wearables, Voice Assistants, Hardware