Last year, Google DeepMind sent gaming values plummeting with the reveal of its Genie 3 global model. Since then, global models have become a priority of AI labs around the world, but Google isn’t resting on its laurels. Researchers have been hard at work on an updated model that improves Genie 3 in almost every way. And if you pay Google enough money, you might be able to use Genie 4 before too long.
It’s easy to look at the release of Genie 4 and see a video game, which is why Genie 3 caused so much consternation in the video game industry. However, DeepMind doesn’t think of things that way. According to Google, global models are a key element in the march towards artificial general intelligence (AGI). A world model creates a virtual world, but it does not require prior rendering environments or programming interactive elements. You simply tell the model what you want, and it generates the experience frame by frame in real time as you explore it.
DeepMind says this technology could create infinite training environments for physical robots or open new avenues of rapid prototyping for creative fields. Of course, this requires the model to accurately simulate what you want. Genie 4 has several improvements that make its simulations more useful, thanks in part to Google’s updated Gemini models.
DeepMind says Genie 4 is better at simulating physics.
When you create a world in Genie 4, it leverages Gemini 3.8 Flash as a guidance agent to break down objectives and evaluate each output frame. This keeps the simulation on track, allowing you to move through different environments without losing the original context. However, DeepMind notes that Genie 4’s worlds only stay coherent for a few minutes before forgetting how things started.
Like Genie 3, the new version supports changing settings on the fly. You can change the weather, add elements, or move the camera, and Gemini will help direct the simulation with each update. This is similar to how reasoning models generate additional internal prompts to focus on the right output tokens.
Gn tech
