Bloomberg reported on July 14 that OpenAI's first consumer device is expected to be a rechargeable AI speaker without a screen. The device reportedly incorporates a camera, sensors, and mechanisms that generate movement, turning ChatGPT into a portable "companion" within the home. It represents an attempt to move voice functions—normally invoked from within a smartphone—into a dedicated device placed in living spaces. What is likely to differentiate the product is not speaker audio quality, but how well it understands its surroundings and the user, and how proactively it can act before being asked.
Ditching the screen, carried from room to room
The device under development reportedly handles smart home control, media playback, and answering questions, and also responds to messages. With a built-in rechargeable battery, it can be carried into the laundry room while doing laundry, or into the kitchen while cooking. When not in use, it can also stay plugged in within a single room. The design appears aimed at a middle ground between a stationary speaker and a mobile device.
It also features a camera and multiple sensors that perceive the surroundings. Additionally, some mechanisms reportedly move on their own. However, this does not mean the device itself moves around the room autonomously. The types of moving parts and how they move remain undisclosed.
Bloomberg reports that OpenAI internally describes this product as a "computer for AI," distinct from conventional smart speakers. Around five products were reportedly under consideration for the hardware business, and this speaker was chosen as the first unit. The product name, price, physical dimensions, and battery life remain undisclosed.
GPT-Live layers "listening" and "speaking"
Supporting the conversational side of the product is reportedly GPT-Live, which OpenAI released on July 8. GPT-Live adopts a full-duplex approach, allowing it to speak while listening to the other party. Moving away from the approach of waiting for the user to finish speaking before responding, it continuously decides—during conversation—whether to keep listening, speak, interject, or call a tool.
When search or complex reasoning becomes necessary, GPT-Live hands off the processing to a separate model while continuing the conversation. At launch, GPT-5.5 handles this behind-the-scenes processing. This design—separating response speed as a conversational partner from the time needed for search and reasoning—anticipates longer exchanges than voice assistants that field questions one at a time.
According to OpenAI, more than 150 million people per week talk to ChatGPT using Voice and Dictation. Since there is already a large existing user base, the new device would not need to teach entirely new conversational habits from scratch. On the other hand, at the time of GPT-Live's release, the feature combining voice with video or screen sharing was not yet supported. How footage captured by the camera connects to conversation won't be clear until the product's implementation is revealed.
The price of "anticipation": access to email and surroundings
What OpenAI is aiming for is not a passive speaker that merely carries out instructed tasks. The device is said to deepen its understanding of the user over time and proactively surface information likely to be needed. To do this, it reportedly also references personal information such as email. Convenience scales with how broadly the device can enter into the context of daily life.
This shifts the center of gravity in product design from camera and microphone performance to how permissions are communicated. On a smartphone, an on-screen indicator shows when the camera is active, and permissions can be toggled per app. A device without a screen must instead signal recording or capture status through light, sound, or physical movement. Family members and visitors also need to be able to immediately recognize that status.
How much of email the device can read, and whether video and audio are processed on-device or sent to the cloud, have not been disclosed. Whether there is a way to pause the anticipatory features, or a method to erase accumulated understanding, will also shape how the product is evaluated. Even if a personable "character" is skillfully crafted, if users cannot control the scope of observation, the case for placing it in the home weakens.
2027 launch target and the Apple lawsuit
The collaboration between OpenAI and Jony Ive came into public view with an official announcement in May 2025. At the time, both parties explained that the collaboration had progressed to concrete design work, and on July 9 of that year, the io Products team formally joined OpenAI. Jony Ive and the design firm LoveFrom remain independent while taking charge of design and creative direction across OpenAI. According to Bloomberg, OpenAI spent $6.5 billion to acquire io Products.
The first unit is reportedly targeted for unveiling sometime in 2026, with a launch planned for 2027. However, that timeline now overlaps with a separate clock ticking in the courts. On July 10, Apple filed a complaint in the U.S. District Court for the Northern District of California against former employees Tang Yew Tan and Chang Liu, an OpenAI-affiliated entity, and io Products, alleging trade secret misappropriation and breach of contract.
Apple is seeking an injunction, which—if granted—could delay OpenAI's hardware sales. This is Apple's claim, and no injunction has been granted yet. OpenAI has countered that it has no interest in other companies' trade secrets and is not aware of any evidence supporting the allegations.
If an announcement does materialize within the year, what's worth confirming before appearance details are how camera and email permissions will be explained, when users can turn things off, and where data is processed. This will be material for judging whether the product can be accepted into the home. Meanwhile, what could directly affect the 2027 launch target is the progress of development and how the injunction ruling in the lawsuit plays out. Whether OpenAI reveals a concrete timeline in an announcement within the year will be the next milestone to watch.