Edge Rewrite
everything Source code View original
// HTMLRewriter · presentation

This page was redesigned at the edge.

Cloudflare fetched the original article and streamed it through HTMLRewriter to apply an entirely new visual system without rebuilding the source page.

// request.cf · coarse context

A page that knows where it met you.

Only coarse request metadata is shown. This demo does not display or persist visitor IP addresses.

Country
US
Cloudflare location
CMH
Connection
HTTP/2
Language
Not provided

Ray ID: a23e81eb7eec3e6e

Jump to content

// Workers AI · dad joke modeWhat did the Genie data model say? "You're in my database now.

Add links
From Wikipedia, the free encyclopedia
(Redirected from Waymo World Model)

Genie is a world model developed by Google DeepMind that generates interactive virtual worlds. It is a spatiotemporal transformer, a type of neural network that models what appears in each frame and how the scene changes over time. First introduced in 2024, Genie was initially trained on video game footage and could only generate two-dimensional environments.

Later versions had higher resolution and increased memory capabilities and could generate three-dimensional environments. Waymo used Genie 3 to create a variant called the Waymo World Model for simulating edge cases for robotaxis. In 2026, Google released Project Genie, a web interface that allows subscribers to use Genie 3. Genie is used in video game design and autonomous vehicles.

Versions

[edit]

The original Genie model, introduced in March 2024, was only capable of generating two-dimensional environments. It generates frames of the world from interactive input and previous frames, and was trained on video game footage.[1] The original Genie generated worlds at only one frame per second.[2]

Genie 2, introduced in December 2024, had the capability of generating three-dimensional environments as well as two-dimensional ones.[2] It was limited to generating 10–20-second worlds from isometric and first-person perspectives. Compared to Oasis, Genie 2 has greater memory and rendering capabilities.[3]: 76  Genie 2 could generate worlds in 360p resolution.[4]

Genie 3, introduced in August 2025, is capable of generating and modifying interactive worlds in real-time. It has a memory of one minute, which is higher than Genie 2's 10–20 seconds.[5] Genie 3 generates interactive worlds at 720p and 24 frames per second,[5] and can produce street environments using Google Street View data.[6]

Technical specifications

[edit]

Genie is a spatiotemporal transformer that models both the structure within individual frames and the changes between frames. It operates in a latent space, an internal mathematical representation that allows text and image prompts to influence the generated environment. The model contains 11 billion parameters.[7]: 301 

Applications

[edit]

Genie was designed for applications in robotics and simulations.[6] It has been used in video game design, including creating imitations of Nintendo's games Super Mario 64 and The Legend of Zelda: Breath of the Wild.[8]

Self-driving cars

[edit]

Waymo adopted Genie 3 to create the Waymo World Model, a system for simulating edge cases faced by robotaxis. Compared with Genie 3, it outputs lidar four times faster and produces more realistic simulations.[9][10] It accepts dashcam or smartphone video as input[11] and provides three forms of simulation control: driving action control, which determines how the model responds to driving input; scene layout control, which specifies the road layout and the behavior of other vehicles; and language control, which specifies environmental conditions.[12][13] It later helped introduce Waymo's robotaxis to 11 cities in the United States.[14]

Project Genie

[edit]

Project Genie is a website that allows AI Ultra subscribers to access Genie 3.[15] It was released on January 29, 2026.[16] It is accessible via Google Labs, an experimental products platform.[17] The site limits world exploration to 60 seconds[18] because Genie 3 is an autoregressive model that requires substantial dedicated compute, making longer sessions too expensive to scale to more users. DeepMind says extending the limit would add relatively little testing value while significantly increasing compute costs.[16]

In generated worlds, the WASD keys are used to move, the arrow keys are used to turn the camera, and the space bar is used to ascend.[16] Project Genie has an integration with Google Street View, allowing it to also generate streets.[14] Following its release, the stock prices of several video game companies declined.[19] Thomas Smith of Fast Company reported that Genie could help users create new types of games in the future, games where users can build custom worlds based on their own preferences.[20]

Jay Peters of The Verge found that that it could generate the worlds of Nintendo games but not worlds containing Disney characters, describing its performance as slow.[8] Smith argued that its limitation of 24 fps was insufficient for practical applications, describing it as "too choppy for real-world use", with modern systems capable of achieving 120 fps at the time of its release.[20] TechCrunch reporter Rebecca Bellan also discovered that model generation was limited, noting that it could not generate content related to nudity or copyrighted material.[16]

References

[edit]
  1. ↑ Orland, Kyle (March 5, 2024). "Google's Genie game maker is what happens when AI watches 30K hrs of video games". Ars Technica. Archived from the original on December 8, 2024.
  2. 1 2 Orland, Kyle (December 6, 2024). "Google's Genie 2 "world model" reveal leaves more questions than answers". Ars Technica. Archived from the original on December 13, 2024.
  3. ↑ June Tay; Lee Cheng Tan; Tai Wei Lib (2026). Ai And Chatgpt : A Techno-legal-socio Analysis. World Scientific Publishing Co. Pte. Ltd. ISBN 9789819821976. LCCN 2025047304.
  4. ↑ Kan, Michael (August 6, 2025). "Google's Genie 3 Hints at a Future Where AI Builds the Video Games We Play". PCMag. Retrieved April 12, 2026.
  5. 1 2 Whitwam, Ryan (August 5, 2025). "DeepMind reveals Genie 3 "world model" that creates real-time interactive simulations". Ars Technica. Archived from the original on March 24, 2026. Retrieved April 15, 2026.
  6. 1 2 Kirchner, Malte (May 25, 2026). "Like 2021 with LLMs: Google researchers on the future of world models". heise online. Retrieved May 27, 2026.
  7. ↑ Heiko Hamann (May 10, 2026). Swarm Robotics. Springer Nature Switzerland. ISBN 9783032105844. Retrieved July 6, 2026.
  8. 1 2 Peters, Jay (January 29, 2026). "Google's AI helped me make bad Nintendo knockoffs". The Verge. Retrieved April 6, 2026.
  9. ↑ Martindale, Jon (February 6, 2026). "Waymo Is Using Google's Genie 3 AI to Practice Handling Tornadoes, Elephants". PCMag. Retrieved April 6, 2026.
  10. ↑ Whitwam, Ryan (February 7, 2026). "Waymo leverages Genie 3 to create a world model for self-driving cars". Ars Technica. Retrieved April 14, 2026.
  11. ↑ Burnett, Steward (February 9, 2026). "Waymo unveils DeepMind-powered world simulation model". Automotive World. Archived from the original on February 10, 2026. Retrieved April 15, 2026.
  12. ↑ Sprigg, Sam (February 12, 2026). "Waymo Launches Generative 3D World Model to Train Self-Driving Cars". Auganix. Archived from the original on April 15, 2026. Retrieved April 15, 2026.
  13. ↑ Hawkins, Andrew (February 6, 2026). "What happens when Waymo runs into a tornado? Or an elephant?". The Verge. Retrieved April 15, 2026.
  14. 1 2 Bellan, Rebecca (May 19, 2026). "Google's Genie world model can now simulate real streets with Street View". TechCrunch. Archived from the original on May 19, 2026. Retrieved May 26, 2026.
  15. ↑ Conditt, Jessica (May 19, 2026). "Project Genie adds Google Street View integration and goes live for global AI Ultra users". Engadget. Retrieved May 27, 2026.
  16. 1 2 3 4 Bellan, Rebecca (January 29, 2026). "I built marshmallow castles in Google's new AI-world generator". TechCrunch. Retrieved April 13, 2026.
  17. ↑ Vigliarolo, Brandon (January 29, 2026). "Google's Project Genie could put even more game developers out of work". The Register. Archived from the original on February 18, 2026. Retrieved April 13, 2026.
  18. ↑ Whitwam, Ryan (January 29, 2026). "Google Project Genie lets you create interactive worlds from a photo or prompt". Ars Technica.
  19. ↑ Pernell, Avalon (January 30, 2026). "Unity, Video Game Stocks Fall as Google's AI Tool Sparks Fears". Bloomberg News. Retrieved April 14, 2026.
  20. 1 2 Smith, Thomas (February 5, 2026). "How Google's new AI could change videogames forever". Fast Company. Retrieved April 6, 2026.

Further reading

[edit]
[edit]
Retrieved from "https://en.wikipedia.org/w/index.php?title=Genie_(world_model)&oldid=1367001693#Self-driving_cars"