(Image source from: Google)
Months after showcasing the Genie 3 AI, Google is now offering access to Project Genie as a test research model. On Thursday, the tech company based in Mountain View announced that certain premium Gemini subscribers can now create AI-driven playable virtual environments in Gemini. This experience utilizes the same Genie 3 model and makes use of Nano Banana alongside Gemini 3, allowing users to build fully interactive worlds based on simple language instructions. It is worth mentioning that Genie 3 was initially revealed in August 2025. In a blog entry, the tech company explained and introduced Project Genie, which is currently open to Google AI Ultra subscribers in the United States, for individuals aged 18 and older. This feature enables them to create, explore, and modify interactive virtual environments. In the US, the Ultra plan costs $249.99 (around Rs. 22,980) per month. The same plan is available in India for Rs. 24,500 a month, but the feature is not offered in that region.
Genie is one of the key initiatives from Google DeepMind. This division has developed three versions of large language models focused on creating virtual environments and claims it contributes to Google's broader goal of artificial general intelligence. Genie 3 is the most sophisticated version, capable of generating both 2D and 3D virtual spaces, while also showing the pathway in real-time as the user navigates and interacts with the environment. This model also ensures realistic physics during interactions in the ever-changing environments. The tech company asserts its use cases extend across robotics, animation modeling, and additional fields. Through Project Genie, Gemini users can input a prompt or upload images to fashion an interactive and evolving setting. They can also create a character and determine how they wish to experience the world. The feature allows for walking, riding, flying, driving, and more. Moreover, the World Sketching feature, which runs on Nano Banana Pro, gives users detailed control over the environment they create.
The World Sketching tool provides a preview of how the environment might appear, allowing tweaks to refine the final output before diving in. Users can choose between first-person or third-person views. This feature also includes a Remix tool, letting users build upon their earlier created environments. Since this is a research prototype, Google cautions that the created worlds may not look entirely realistic or closely follow the provided prompts and images. They also noted that characters might sometimes be difficult to control or show delayed responses while exploring. Additionally, all world creations are constrained to 60 seconds of exploration time.
