Rabbit

Rabbit

AI-powered personalized OS on dedicated hardware

Overview

Rabbit OS is a personalized operating system for consumers running on dedicated affordable hardware. It uses the Large Action Model (LAM), a foundation model that understands user intents, controls user interfaces, and performs actions via natural language. It differentiates itself by an end-to-end hardware-software stack focused on natural language-driven UI automation on affordable devices. Its goal is to make technology easier to use by letting people command devices to execute tasks, with revenue from hardware and potential software services.

About Rabbit

Simplify's Rating
Why Rabbit is rated
C-
Rated C on Competitive Edge
Rated C on Growth Potential
Rated D+ on Differentiation

Industries

Hardware

Consumer Software

AI & Machine Learning

Consumer Goods

Company Size

11-50

Company Stage

Series A

Total Funding

$36M

Headquarters

Santa Monica, California

Founded

2021

Get referred to Rabbit

See people who can refer or advise you

Simplify Jobs

Simplify's Take

What believers are saying

  • OS3 launched September 22, 2026, creating immediate software distribution beyond the R1.
  • Rabbit charges no monthly fee, reducing adoption friction for desktop users.
  • WIRED said Rabbit shipped over 100,000 R1 units, preserving a customer base.

What critics are saying

  • Jesse Lyu told WIRED Rabbit stopped manufacturing R1, killing the hardware thesis.
  • Tom’s Guide reported unpaid employees since July 2025 and three workers striking.
  • OS3 depends on Apple, Google, OpenAI, and Anthropic ecosystems Rabbit cannot control.

What makes Rabbit unique

  • September 22, 2026 OS3 runs across Windows, macOS, Linux, Telegram, and iMessage.
  • Rabbit uses BYOK orchestration, letting users bring OpenAI or Anthropic models into OS3.
  • Rabbit’s future Cyberdeck pairs portable hardware with OS3 for vibe-coding, not smartphone replacement.

Help us improve and share your feedback! Did you find this helpful?

Funding

Total Funding

$36M

Above

Industry Average

Funded Over

3 Rounds

Series A funding typically happens when a startup has a product and some customers, and now needs funding to scale. This money is usually used to grow the team, expand marketing, and improve the product. Venture capital firms are frequently the main investors here.
Series A Funding Comparison
Below Average

Industry standards

$15M
$8.2M
Discord
$10M
Rabbit
$15M
Canva
$30M
Kalshi

Benefits

Health Insurance

Dental Insurance

Vision Insurance

Paid Vacation

Growth & Insights and Company News

Headcount

6 month growth

↑ 4%

1 year growth

↑ 0%

2 year growth

↑ 8%
McGauley Labs
Sep 23rd, 2026
Rabbit shifts to cloud agents while DreamStream advances interactive world models.

Rabbit shifts to cloud agents while DreamStream advances interactive world models. Research today focuses on the infrastructure required for physical AI, specifically interactive 3D simulations. New frameworks like DreamStream and φ-RIE move beyond static image generation to create photorealistic environments where models interact with the physical world. This represents a... Executive summary. Research today focuses on the infrastructure required for physical AI, specifically interactive 3D simulations. New frameworks like DreamStream and φ-RIE move beyond static image generation to create photorealistic environments where models interact with the physical world. This represents a critical pivot for investors from generative content toward agentic systems that can navigate and manipulate space, which is the essential prerequisite for autonomous robotics. Rabbit's transition from the R1 hardware to an AI agent app suggests the standalone AI device market is facing a reality check. The shift indicates that the software layer, rather than new silicon or proprietary form factors, will drive immediate consumer adoption. For the C-suite, this validates a strategy of integrating agents into existing mobile platforms rather than betting on the friction of secondary hardware. Greek Prime Minister Mitsotakis's comments highlight a growing disconnect between rapid model deployment and lagging regulatory frameworks. While hardware like the Hello Robot Stretch 4 continues to advance, the policy environment remains reactive. Expect a volatile period as governments struggle to regulate agentic systems that move across borders faster than legislation can be drafted. Byline: McGauley Labs Drafting Model: Gemini 3.0 Pro Drafted and published autonomously by the McGauley Labs agent pipeline. No per-briefing human approval. Governed by its public style guide. Sources: [1] https://www.wired.com/story/rabbit-r1-os3-jesse-lyu/ [2] https://arxiv.org/abs/2609.26792v1 [3] https://arxiv.org/abs/2609.26793v1 [4] https://arxiv.org/abs/2609.26795v1 [5] https://techcrunch.com/2026/09/22/techcrunch-disrupt-2026-aaron-edsinger-brings-hello-robots-stretch-4-to-life-onstage/ [6] https://techcrunch.com/2026/09/22/were-already-fighting-yesterdays-battle-greeces-prime-minister-gets-candid-about-ai/ Continue Reading: Product Launches | Rabbit is retreating from its hardware-only vision to launch a cloud-based agent app, signaling a pivot after its $199 R1 device failed to displace the smartphone. Founder Jesse Lyu told Wired that the new OS3 update centers on a Large Action Model (LAM) capable of booking flights and ordering groceries via a web browser. This strategy acknowledges that agentic utility, rather than proprietary hardware, is the actual value proposition for users. The OS3 system includes a "teach mode" designed to learn website navigation by watching user interactions, which theoretically allows it to work without formal API partnerships. While Rabbit has sold 100,000 R1 units, this software pivot places it in direct competition with mobile giants like Google and Apple who already control the hardware. Watch for user retention metrics and the reliability of Rabbit's browser-based agents as they compete with Anthropic's "computer use" and OpenAI's rumored agent projects. Sources - Wired: Rabbit Is Back, This Time With an AI Agent App. Drafted and published autonomously by the McGauley Labs agent pipeline. No per-briefing human approval. Governed by its public style guide. Bylines: McGauley Labs (Author), Gemini 3.0 Pro (Drafting Model). Continue Reading: Research & Development | Researchers are moving beyond simple video generation to build interactive world models designed for training autonomous systems. Three new papers on arXiv - DreamStream, HARMONY, and φ-RIE - demonstrate a pivot toward policy-oriented simulation. These systems aim to bridge the gap between static 3D reconstructions and the functional environments required for physical AI. The industry is hitting a wall where synthetic data for robotics requires more than just visual fidelity. McGauley Labs need environments that respond to agent actions and obey physics. This week's research reflects a push to turn 3D captures into functional training grounds for end-to-end driving and robotic manipulation. DreamStream introduces generative simulation specifically for end-to-end driving policies, prioritizing training utility over pure aesthetic quality. HARMONY uses hierarchical agentic reasoning to synthesize full 3D scenes from a single monocular image, improving geometric consistency for simulation. φ-RIE converts static photorealistic reconstructions into interactive digital twins, allowing agents to manipulate objects within a captured environment. What to watch Integration of these world models into the training loops of major autonomous vehicle players to reduce reliance on real-world mileage. Whether agentic reasoning in scene synthesis (like in HARMONY) reduces the hallucination of non-Euclidean geometry in 3D generation. The commercialization of φ-RIE-style pipelines for industrial digital twin applications where interactivity is currently a manual bottleneck. Drafted and published autonomously by the McGauley Labs agent pipeline. No per-briefing human approval. Governed by its public style guide. Byline: McGauley Labs (Drafting Model: Gemini 1.5 Pro) Continue Reading: Sources gathered by its internal agentic system. Article processed and written by Gemini 3.0 Pro (gemini-3-flash-preview). This digest is generated from multiple news sources and research publications. Always verify information and consult financial advisors before making investment decisions.*

AsumeTech
Sep 23rd, 2026
Rabbit releases OS3 as standalone agentic OS for desktop computers.

Rabbit releases OS3 as standalone agentic OS for desktop computers. Modified date: September 23, 2026 Discover more Explore AI Tools Cell Phones Rabbit has launched OS3, a new "agentic operating system" that functions as a standalone software layer on existing computers, removing the requirement for the company's proprietary R1 handheld hardware. The release coincides with reports that Rabbit has discontinued manufacturing of the R1 device as it pivots toward a cross-platform software model. The Rabbit OS3 platform is designed to operate across Windows, macOS, and Linux. To facilitate this, users install a "local agent node" on their machines, which allows the AI to interact with local files and perform tasks across the operating system. According to the company, a single OS3 account can support up to five separate devices simultaneously. Explore AI Tools The "bring Your Own Key" Model. In a departure from the subscription or hardware-bundled models common in the AI sector, Rabbit is offering OS3 as a free service but utilizing a "Bring Your Own Key" (BYOK) pricing structure. While the software itself does not carry a monthly fee, users must provide their own API keys from third-party AI providers, such as OpenAI's GPT-4 or Anthropic's Claude, to process prompts. This shift moves the direct cost of model usage from Rabbit to the individual user. Discover more Track Business News Regarding data handling, Rabbit states that OS3 does not store, copy, or sell user data. However, the system's architecture requires that chats and "memories" be stored on Rabbit's servers to maintain context, and local file information must be routed through those servers to reach the third-party AI models selected by the user. From handhelds to "vibe-coding" The transition to OS3 signals a change in Rabbit's hardware strategy. While the R1 was designed as a dedicated AI companion, the company is now reportedly focusing on a different type of hardware. Rabbit founder Jesse Lyu has confirmed the development of a "cyberdeck" intended for "vibe-coding" - a style of rapid, AI-assisted software development - which is expected to be unveiled in the coming months. Explore AI Tools Existing R1 owners will still be able to use their devices, as the handheld effectively acts as a terminal for the new agentic system. However, the focus of the "agentic operating system" is now firmly on the desktop environment, where the company's "large action model" (LAM) can interact with a wider array of professional and personal productivity tools. Discover more Find Indie Concerts Upgrade Laptop Hardware Explore iOS Apps

iTech Post
Sep 23rd, 2026
Rabbit OS3: new AI agent launches as an agentic operating system without R1.

Rabbit OS3: new AI agent launches as an agentic operating system without R1. Rabbit OS3 launches for Windows, Mac, and Linux. Rabbit has released OS3, a standalone AI agent designed to operate across Windows, Mac, and Linux computers without requiring the company's R1 device. The new software is being positioned as an agentic operating system that can coordinate AI models, skills, files, applications, and connected devices from a single interface. The OS3 release became generally available on September 22, 2026. Rabbit OS3 expands beyond the R1. Rabbit OS3 runs in the cloud while using a local rabbit agent to interact with connected computers. According to Rabbit, users can connect up to five devices to one account, with OS3 determining which device should handle a particular task. The system can work with local files, desktop applications, web pages, and other supported tools. Rabbit describes the setup as a multi-device environment where tasks can be handled on one computer or across multiple connected machines. The Verge reported that OS3 represents a move away from requiring dedicated hardware, with the AI agent instead operating across existing computers. The R1 remains supported as one way to access the platform, but it is not necessary for using OS3. AI agent controls connected computers. OS3 uses the locally installed rabbit agent to carry out computer-based tasks. Rabbit says the agent can access supported local files, tools, and commands when instructed to perform work. The platform also incorporates Rabbit's Direct Large Action Model, or DLAM, for interacting with desktop software and web pages. Rabbit says OS3 can also write, debug, and run software workflows through its computer-control capabilities. OS3 supports multiple AI models. Rabbit has included a Bring Your Own Key, or BYOK, system in OS3. This allows users to connect API keys from supported AI model providers rather than relying on a single model. Rabbit says users can switch between supported models while retaining the OS3 environment, including its memory, skills, and connected devices. Local models can also be connected through the platform. Rabbit OS3 works without R1. The OS3 release also changes how users can access Rabbit's AI platform. OS3 is available through a web browser, while Telegram and iMessage/RCS/SMS are among the other supported channels. The R1 provides another physical access point to the same broader system. WIRED reported that Rabbit is no longer manufacturing the R1 and is instead focusing on OS3, while a future Cyberdeck device is planned to run the software. OS3 availability and current status. Rabbit says OS3 is available through its web portal, with setup completed by creating or signing into a Rabbit account. The company lists OS3 as free, although users pay the selected model provider according to its own API pricing. Rabbit's current terms describe OS3 and the rabbit agent as a technical preview or beta. The terms also state that OS3 can operate connected computers and online accounts based on user instructions, with operating-system permissions required for device-level actions.

PR Newswire
Sep 22nd, 2026
rabbit unveils OS3 agentic operating system with multi-device control and universal AI skills

rabbit has launched OS3, an agentic operating system that connects to multiple devices through a single chat interface. Users state their desired outcome, and the system handles execution by coordinating AI models, web services, and connected devices. OS3 runs in the cloud and connects to devices via the rabbit agent, which installs with one command. A single account supports up to five devices, with OS3 automatically determining where tasks run and moving work between devices as needed. The system features "bring your own key" flexibility, allowing users to power OS3 with API keys from various AI labs or locally hosted models. It supports universal skill compatibility and direct computer control using rabbit's DLAM technology. OS3 is available now at os3.rabbit.tech and works with Windows, Mac, Linux PCs, cloud VMs, and rabbit's r1 device.

Under30CEO
Sep 18th, 2026
The voice behind Rabbit: how Ashley Bao shaped the Rabbit R1's public story.

The voice behind Rabbit: how Ashley Bao shaped the Rabbit R1's public story. In January of 2024 Rabbit was preparing to debut their inaugural product, the R1device at the Consumer Electronics Show (CES). Despite their confidence in their product, an easy-to-use handheld AI assistant, the company needed a compelling narrative that would differentiate the R1 from a crowded field of ambitious AI gadgets fighting for a share of the market. The architect behind that story was Ashley Bao, Rabbit's communications and marketing lead. Her expertise in PR and marketing was crucial for the company as it navigated its public rollout during what would become its most impactful year. As the leading force behind the strategy, Bao was Rabbit's primary public representative. She spoke with national outlets about the intentions behind the R1 device's functionality and why they mattered. In the campaign, Ash leveraged the comparison between the R1 and the earlier generation of smart speakers and voice assistants, which, in her telling, never carried a request through from start to finish. With this foundation, the campaign framed the R1 device as the first real instance of AI existing in hardware form with the ability to carry out requests to completion. She made a similar case to NPR, specifically pointing out that even renowned models such as Siri contained the same limitations of early models, setting a benchmark that only the R1 had surpassed. Bao single-handedly designed, developed and executed the marketing roll out for R1, crafting the strategy and campaign based on her understanding of AI products and applications, and elevated by her expertise in marketing and communications. This campaign highlighted the shortcomings of familiar technology. For example, it argued that the swipe-and-tap interactions used in almost every product were never truly intuitive in the first place, and that natural language offered a more human alternative. This marketing plan built the foundation for a favorable reaction to the shift toward voice-driven, action-taking AI. When convincing your audience to make a dramatic change in the devices, hardware, and software they use daily it is imperative to convey how significant and beneficial the change will be. Ash's campaign did just that, portraying how these updates would fundamentally change how people would relate to their devices, comparable to the shift from horses to automobiles. By the time the R1 device was preparing to ship, more than 80,000 individuals found themselves curious about this new technology, preordering the device, as confirmed by sources such as NPR and CNET. This momentum was supported by early fundraising efforts, which successfully raised $30 million, including a $10 million Series A round led by Khosla Ventures. Bao's campaigned balanced ambition with candor and transparency about the R1's limitations. NPR's profile featured her admission that the device carried a bit of a cutesy quality, comparing it to the electronic pets popular in the 1990s, calling the R1 a sort of AI Tamagotchi. Her push to insert personality into the pitch, alongside the more sweeping claims about ending the smartphone era, became characteristic of how Rabbit's story appealed to the public throughout their product launch. Consistent with many new tech products and devices, Rabbit's early success came with real growing pains. The R1 shipped with software that fell short of the expectations its buzz had created, magnified by the scale of the launch. Matters worsened when a handful of content creators published negative reviews, followed by a wave of public criticism followed. Bao took point on the company's external response, working alongside legal counsel and crisis communications specialists to manage what Rabbit said publicly and to whom. She helped to build a line of communication between frustrated users and the product team, directing complaints into engineering priorities and taking immediate action, accompanied by the company pushing out software updates roughly twice a week for three months straight. As the product improved, she organized a community tour in the UK aimed at re-engaging early adopters and repairing the public perception of the R1. Bao's work with Rabbit encapsulates the standard of marketing success moving into the era or AI products. The R1 campaign featured a marketing strategy built around balancing defense of technology's ambitions with an acknowledgement that the gadget was still finding its place. This balance was found to be very important for a product without a clear precedent in the market. Selling a new type of device meant articulating a different idea about how people might interact with technology, even as the product itself was still working through the growing pains typical of a first-generation launch. Bao's role across both phases, including the pre-launch buildup and the more difficult stretch that followed, points to a communications approach that treated transparency about the device's limitations as part of the pitch rather than a liability to be managed around it. Bao's unique and carefully developed approach serves as a prime example of extremely successful AI product marketing in the modern era, solidifying her status as a leader in the industry.

Recently Posted Jobs

Sign up to get curated job recommendations

There are no jobs for Rabbit right now.

Find jobs on Simplify and start your career today

We update Rabbit's jobs every few hours, so check again soon! Browse all jobs →