Rabbit OS3 is the new “agentic operating system” from startup Rabbit, which is set to change the face of AI agents — and this time you don’t need the controversial R1 hardware to use it. The company has announced a standalone AI that runs in the cloud but works locally on Windows, Mac , and Linux, paving the way for a new era in the use of AI tools.
Rabbit made a lot of noise with the launch of the R1, a device that promised to revolutionize the way we interact with artificial intelligence. However, the reality turned out to be much more mundane: the R1 was heavily criticized for its limited functionality, bugs, and general inability to meet the high expectations that had been created. Now, the company is attempting a new beginning with Rabbit OS3, decoupling its software from the hardware.
According to The Verge, OS3 allows users to connect up to five devices to one account, along with their preferred AI models. The system automatically decides which devices, files, apps, and AI models it needs to complete a task — without requiring manual intervention from the user.
See also: Can an AI agent become a hacker without a human to guide it?
Rabbit OS3: How the new agentic system works
Rabbit OS3 is fundamentally different from traditional AI assistants. Instead of simply answering questions, it acts as a true AI agent that takes action: opening applications, managing files, coordinating different AI models, and performing complex multi-step tasks. This “agentic” approach represents the next generation of AI tools, where the system does not wait for instructions for each step but acts on its own to achieve a goal.
Access to OS3 is extremely flexible. Users can use it through a dedicated desktop site, through messaging apps like Telegram or iMessage, and of course through the R1 for those who already own it. This multi-channel approach shows that Rabbit wants to reach as wide an audience as possible, without depending on hardware sales.
One of the most interesting features of Rabbit OS3 is its support for multiple AI models. The user can choose the models they prefer — whether they are models from OpenAI, Anthropic, Google , or other providers — and OS3 will automatically tune them according to the needs of each task. This gives users great flexibility and avoids being “locked-in” to a single AI provider.
Rabbit OS3 and the future of R1: End of an era for hardware?
Rabbit founder Jesse Lyu revealed to Wired that the company has stopped producing new R1 devices. This development is not surprising, given the disappointing reception to the product. However, Rabbit is not abandoning the AI hardware space entirely — instead, it is turning its attention to a new, more ambitious project.
See also: Revolver Rabbit gang registers 500,000 domains for malware campaigns
It’s a “cyberdeck” designed around the philosophy of vibe coding — a trend that’s gaining traction in the tech community that involves creating code and apps with the help of AI, without traditional programming knowledge. The new hardware is expected to be released “within months,” according to Wired, and will run OS3 as its primary operating system. If Rabbit can deliver on its promises, it could be a much more complete proposition than the R1.
Rabbit’s strategic shift — from a standalone hardware product to a cross-platform software ecosystem — reflects a broader trend in the AI space. Companies like Humane with its AI Pin have also faced serious difficulties in trying to sell dedicated AI hardware, proving that consumers are not yet ready to replace their smartphones with specialized devices.
Rabbit OS3 and privacy
A critical issue for any AI agent is privacy and security. Rabbit claims that OS3 does not store, copy, use, or sell user data. However, there is one important caveat: the company admits that it will keep conversations and the “memories” that result from them on its servers. In addition, third-party AI providers connected to OS3 will handle the data according to their own privacy policies.
This means that users need to be aware that their data may be processed by multiple providers, each with their own rules. At a time when awareness around data protection is growing — particularly in Europe with GDPR — this complexity can be a barrier for many users.
See also: Rabbit Announces R1: The Next Generation of Autonomous Artificial Intelligence Devices
To protect yourself from risks that may be associated with the use of AI agents, such as Rabbit OS3, it is recommended that you carefully read the privacy policies of both Rabbit and the third-party AI providers you connect to your account. Avoid sharing sensitive personal information or business data through any cloud-based AI system, and regularly review which applications and devices have access to your account.
Overall, Rabbit OS3 represents an interesting development in the field of AI agents, freeing the capabilities of artificial intelligence from the constraints of dedicated hardware. If the company manages to deliver on its promises — especially in terms of data security and system reliability — OS3 could be a truly useful platform for users who want to harness the power of multiple AI models from a central interface. The coming period, with the release of the new cyberdeck, will prove whether Rabbit has truly learned from the mistakes of R1.
