Google announced on Thursday that it's expanding access to its experimental research prototype, Genie 3, to its AI Ultra subscribers. The experiment allows people to generate 3D worlds and navigate through them by using text prompts.
Google gave a glimpse of Genie 3 in August, and until now, access has been limited to its Trusted Testers program. Now, Genie 3 will be available on the highest tier of Google's AI subscription plan, limited to people 18 and older.
Google calls Genie 3 a general-purpose world model that lets people generate dynamic worlds, which can be navigated in real time. Seeing the demos in action, it's like creating a video game on the fly. In fact, several video game stocks sank in response to Google's announcement.
Powered by Gemini, Nano Banana Pro and Veo 3, Genie 3 is backed by some of the best of Google's AI products.
Don't miss any of our unbiased tech content and lab-based reviews. Add CNET as a preferred Google source.
Google breaks down Genie 3's functions in three ways.
World sketching: Using text prompts and uploaded images, people can craft a world as they see fit, along with the character that will navigate it. You can choose how the character navigates the world, whether that be walking, flying, or something else, and you can even select the perspective of the character, like first-person or third-person views.
World exploration: With your world built, Genie 3 will generate the path ahead in real time based on your interactions and movements. You can also adjust the camera view and angle as you navigate through the world.
World remixing: Google will allow people to remix worlds by building on top of the prompts that created them. Genie 3 offers a curated gallery of worlds to choose from. You can then download videos from your world explorations.
Since Genie 3 is still an early experimental release, it's not without a few limitations.
... continue reading