DeepMind previews Genie 4 world model guided by Gemini 3.8 Flash

Source – Since then, world models have become a major focus 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 on Genie 3 in almost every way, and paying customers may soon be able to use Genie 4.

DeepMind frames world models as a key element in the march toward artificial general intelligence (AGI). A world model creates a virtual world without pre-rendered environments or programmed interactive elements; users simply describe what they want and the model generates the experience frame by frame in real time.

According to DeepMind, the technology could create infinite training environments for physical robots or unlock new avenues of rapid prototyping for creative fields, provided the model accurately simulates the desired scenario. Genie 4’s improvements stem in part from Google’s updated Gemini models.

When a world is prompted in Genie 4, it leverages Gemini 3.8 Flash as a guiding agent to break down goals and evaluate each output frame, helping keep the simulation on track as users move through varying environments. DeepMind notes, however, that Genie 4 worlds remain consistent for only a couple of minutes before they forget how things started.

Like Genie 3, the new version supports changing parameters mid-stream. Users can alter weather, add elements, or move the camera, and Gemini assists in steering the simulation with each update, similar to how reasoning models generate internal prompts to refine output tokens.