In an exclusive interview with Tom’s Guide, Rabbit CEO Jesse Lyu has confirmed that work has begun on a next-generation sequel to the Rabbit R1, and teased that it could be a three-in-one device set to launch in 2026.
This chance to speak with Lyu comes a few weeks after the launch of RabbitOS 2. A re-review of the Rabbit R1 is coming next week (spoiler alert: it’s really rather good now) alongside the full interview with Lyu, but I’m too impatient to just sit on these exciting details!
Whatever you thought about the AI hardware war after the initial debut of the Rabbit R1, the implosion of the Humane AI Pin or the protests surrounding the Friend pendant, it’s far from over. Rabbit has its sights set on taking the fight to OpenAI next year.
Editor’s note: These answers have been edited for brevity and clarity
Allow me to reintroduce myself
Watch On
So after a big conversation about where Rabbit stands right now, the question was raised about when the company is going to do “gen two.” And the answer, seemingly, is 2026.
“We made a responsible decision this year not to rush into the new form factor. That’s why we spent every bit of our capacity to work on the R1 software,” Lyu commented.
“But we do have a couple of the prototypes exploring different form factors. 2026 is gonna be interesting because OpenAI will join the competition, and I assume there’s other startups that will also join the competition.”
This was an especially interesting point, because AI hardware (and our attitude towards AI hardware) has changed over the last two years. Yes, it’s still being made, but there’s an increasing amount of skepticism about it — namely in the vein of asking “why does it need to be an extra device?” Jesse understands this first-hand.
“I agree that you shouldn’t carry two devices. No one wants to carry two devices. In fact, we were not expecting you to carry your R1 with you, even though we allow a sim card to make it mobile — even though a lot of our fans take them everywhere,” he adds.
“I think the general thoughts in our product team right now is that we don’t think the agentic phone is ready yet. Just because there’s a lot of actual serious stuff that you will never trust an AI agent to do, like banking, for example.”
Lyu believes it will take “maybe a year, to a year and a half” for the likes of the iPhone to be an agentic phone, which gives Rabbit “a good amount of time that we can at least do one generation of device.”
So what is this device?
Could it be an iterative upgrade over the R1? Is it the wristband that was teased at the NYC pickup party back in 2023? Actually, it’s sounding like Lyu is most excited about something completely new.
“Right now, we have some candidates we’re internally debating. It will be a hundred percent Rabbit culture-like, so don’t expect a very generic look or generic way of interacting with it. Actually, one of the ideas is a three-in-one device,” he confirmed.
“I cannot share more details, but it’s not just spamming features. It has some other components that make it actually a three-in-one device. So that is probably one of the ideas we’ve been exploring the furthest so far. We have a couple of other ideas, but it would be nice to see what the next form factor will be that is different from what we’re seeing right now.”
Two things that we know it won’t be are smart glasses or a pendant like the Friend, as Jesse had some gripes about these. “We were debating internally whether there’s a generic fit [for glasses], but we don’t think there is. That’s one of the reasons that we haven’t jumped into this smart glasses war,” Lyu said.
“As for the pendants like the Friend, I think they were not at our standard hardware-wise, just by looking at the build quality and the design of it.”
Predicting the three-in-one device
Now, what would make a three-in-one AI device? I fired up the crystal ball and started predicting with the AI team here at Tom’s Guide. Because in my mind, two are clear: AI generation and a chatbot. These are two things that the Rabbit R1 thrives on at the moment.
The third one is a little more interesting, and I think that will be the next generation of agentic capabilities. You’re already seeing some of this at play with Creations and the LAM playground in Rabbit R1, but this could go deeper and become more accessible to a wider audience.
Think like a continuously running assistant agent, which can monitor context and surface suggestions pre-emptively, better capabilities in multi-step workflows, multi-modal agentism (using the camera to snap a gig poster and ask it to buy a ticket for you), or even customizable, personalized agents with different personalities for specific tasks.
Of course, this should all come with higher guardrails, but it certainly is an exciting prospect!
Regardless, one thing is abundantly clear — the Rabbit R1 will continue to get updates for a long time to come. “I don’t want to have people thinking, ‘oh, they’re gonna stop supporting R1 soon.” No, we’re still working on R1 generational software, and we do have new plans,” Lyu confirmed.
Follow Tom’s Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button!