OutYet reporting
Genie 4 reaches subscribers, with developer access still uncertain
Google has opened its interactive world model in Project Genie to eligible AI Ultra subscribers. The update adds live scene editing and audio, while leaving broader access and API availability unresolved.
Google is rolling Genie 4 out to eligible Google AI Ultra subscribers through Project Genie, a browser-based experimental experience for creating and exploring interactive worlds. Users can prompt changes while moving through a world, including changes to weather or characters, and the model generates audio that adapts to the scene. Google also added a conversational creation agent and a gallery where people can share and remix worlds. This is a meaningful step from a research preview toward hands-on use, though access remains limited to a paid subscriber group. [Google’s November 18 update](https://blog.google/innovation-and-ai/models-and-research/google-deepmind/genie-4-now-available-project-genie/) describes the rollout and features.
The timing matters because Google DeepMind introduced Genie 4 on October 27 as a research preview for trusted testers, with future access planned for AI Ultra subscribers and early-access partners using the Gemini API and Google AI Studio. The November update confirms the subscriber path has begun, but does not say that API or AI Studio access is generally available. Google says AI Pro members can join a waitlist. So developers who want to build on the model should treat the earlier API language as a planned access route, not evidence that they can use it today. [Google’s October preview post](https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation/) and [the current availability update](https://blog.google/innovation-and-ai/models-and-research/google-deepmind/genie-4-now-available-project-genie/) establish that distinction.
The change is also a product shift from the October preview, which emphasized real-time simulation and natively generated sound, toward an interactive creation workflow for subscribers. Google says Genie 4 offers its highest-fidelity visuals and physics, while Genie 3 remains available for longer sessions and panorama-seeded spaces. That makes the choice between versions practical: Genie 4 is positioned for richer, editable worlds with audio, while Genie 3 retains features Google says matter for longer or panorama-based sessions. These are provider descriptions, not an independent head-to-head evaluation. [Google’s October preview post](https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation/) and [its November update](https://blog.google/innovation-and-ai/models-and-research/google-deepmind/genie-4-now-available-project-genie/) describe the models and available options.
For technical users, the confirmed value is access to a world-building prototype that supports prompt-driven changes during a session, rather than a generally available model endpoint. Google has not published API availability details or pricing in this update, and its October post described developer access as a future step. Ars Technica likewise reported around the preview that Gemini API rollout was planned and pricing was not yet available. That leaves important questions about reproducibility, integration, cost, and how well generated worlds preserve consistent behavior unanswered. The subscriber rollout gives users a way to explore the experience, but it does not yet establish that Genie 4 is ready for production development. [Google’s November update](https://blog.google/innovation-and-ai/models-and-research/google-deepmind/genie-4-now-available-project-genie/), [Google’s October preview post](https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation/), and [Ars Technica’s preview coverage](https://arstechnica.com/google/2026/10/deepminds-genie-4-world-model-adds-audio-and-more-realistic-people-to-simulations/) support these access and limitation points.
Related models
Sources
- Genie 4 is now available · Google
- Genie 4: A new generation of world simulation · Google DeepMind
- DeepMind unveils Genie 4 world model promising more realistic physics and humans · Ars Technica